update source code

Demo/Reproduce_Scripts/demo_breast.py

@@ -19,7 +19,7 @@
 from sklearn.metrics import adjusted_rand_score
 from sklearn.cluster import SpectralClustering
 from Space.cons_func import get_results, get_domains
-from Space.utils import calculate_location_adj, plot_results_ari, get_bool_martix
+from Space.utils import calculate_location_adj, plot_results_ari, get_bool_martix,plot_ari_with_removal
 
 adata = sc.read_visium(
     path="./Data/V1_Breast_Cancer_Block_A_Section_1", count_file="filtered_feature_bc_matrix.h5"
@@ -80,9 +80,11 @@
 mul_reults = mul_reults.drop('ground_truth', axis=1)"""
 
 
-plot_results_ari(mul_reults)
-mul_reults = mul_reults.drop("SpaceFlow", axis=1)
-mul_reults = mul_reults.drop("MENDER", axis=1)
+mul_reults = plot_ari_with_removal(mul_reults, 2)
+
+# plot_results_ari(mul_reults)
+# mul_reults = mul_reults.drop("SpaceFlow", axis=1)
+# mul_reults = mul_reults.drop("MENDER", axis=1)
 
 pos_similarity = calculate_location_adj(adata.obsm["spatial"], l=123)
 
@@ -98,8 +100,18 @@
     learning_rate=learning_rate,
 )
 con_martix = model.train()
-sc = SpectralClustering(n_clusters=k, affinity="precomputed", random_state=666)
-labels = sc.fit_predict(con_martix)
-adata.obs["consensus"] = labels
+sClustering = SpectralClustering(n_clusters=k, affinity="precomputed", random_state=666)
+labels = sClustering.fit_predict(con_martix)
+adata.obs["Space"] = labels
+adata.obs["Space"] = adata.obs["Space"].astype('str')
 ari = adjusted_rand_score(labels, gt.values)
 print(ari)
+
+sc.pl.spatial(adata, color="Space", title='Space (ARI=%.2f)'%ari)
+
+sc.tl.rank_genes_groups(adata, 'Space', groups=["2"], reference="0", method='wilcoxon')
+sc.pl.rank_genes_groups(adata, groups=['2'], n_genes=20)
+
+sc.pl.rank_genes_groups_violin(adata, groups='2', n_genes=8)
+
+sc.pl.violin(adata, ['PBX1', 'KRT18', 'COX6C'], groupby='Space')

Demo/Reproduce_Scripts/demo_hip.py

@@ -4,7 +4,7 @@
 
 os.environ["R_HOME"] = "/home/zw/software/miniforge-pypy3/envs/space/lib/R"
 import Space
-
+slide_id = "19"
 from Space.cons_func import (
     run_GraphST,
     run_Leiden,
@@ -20,10 +20,7 @@
 from sklearn.metrics import adjusted_rand_score
 from sklearn.cluster import SpectralClustering
 from Space.cons_func import get_results, get_domains
-from Space.utils import calculate_location_adj, plot_results_ari, get_bool_martix
-
-# change slice
-slide_id = "19"
+from Space.utils import calculate_location_adj, plot_results_ari, get_bool_martix,plot_ari_with_removal
 
 adata = sc.read(f"./Data/Mouse_hippocampus_MERFISH/hip_adata-0.{slide_id}.h5ad")
 adata.var_names_make_unique()
@@ -74,6 +71,9 @@
 mul_reults = mul_reults.drop("MENDER", axis=1)
 mul_reults = mul_reults.drop("SpaGCN", axis=1)
 mul_reults = mul_reults.drop("stLearn", axis=1)
+
+
+# mul_reults = plot_ari_with_removal(mul_reults,8)
 pos_similarity = calculate_location_adj(adata.obsm["spatial"], l=123)
 
 model = Space.Space(
@@ -93,3 +93,5 @@
 adata.obs["consensus"] = labels
 ari = adjusted_rand_score(labels, adata.obs["ground_truth"].values)
 print(ari)
+
+

Demo/Reproduce_Scripts/demo_pa.py

@@ -19,10 +19,11 @@
 from sklearn.metrics import adjusted_rand_score
 from sklearn.cluster import SpectralClustering
 from Space.cons_func import get_results, get_domains
-from Space.utils import calculate_location_adj, plot_results_ari, get_bool_martix
+from Space.utils import calculate_location_adj, plot_results_ari, get_bool_martix,plot_ari_with_removal
 
 
-slide_id = "5"
+slide_id = "14"
+# slide_id = "5"
 adata = sc.read("./Data/SRARmap_pa/MS_raw_Dataset11_Dataset11_data.h5ad")
 adata = adata[adata.obs["slice_id"] == f"BZ{slide_id}", :]
 adata.var_names_make_unique()
@@ -63,12 +64,18 @@
 """collected_results = get_results(func_dict, use_multithreading=True, monitor_performance=True)
 mul_reults = get_domains(adata,collected_results,gt)
 mul_reults = mul_reults.drop('ground_truth', axis=1)"""
-plot_results_ari(mul_reults)
+# plot_results_ari(mul_reults)
+# mul_reults = mul_reults.drop('Leiden', axis=1)
+# mul_reults = mul_reults.drop('SpaGCN', axis=1)
+# mul_reults = mul_reults.drop('stLearn', axis=1)
+# mul_reults = mul_reults.drop('SCANPY', axis=1)
+
+
+mul_reults = plot_ari_with_removal(mul_reults,4)
+
+
 pos_similarity = calculate_location_adj(adata.obsm["spatial"], l=123)
-mul_reults = mul_reults.drop('Leiden', axis=1)
-mul_reults = mul_reults.drop('SpaGCN', axis=1)
-mul_reults = mul_reults.drop('stLearn', axis=1)
-mul_reults = mul_reults.drop('SCANPY', axis=1)
+
 model = Space.Space(
     get_bool_martix(mul_reults),
     pos_similarity,
@@ -80,6 +87,7 @@
     beta=1,
     learning_rate=learning_rate,
 )
+
 con_martix = model.train()
 sc = SpectralClustering(n_clusters=k, affinity="precomputed", random_state=666)
 labels = sc.fit_predict(con_martix)

Demo/Reproduce_Scripts/demo_seq.py

@@ -19,14 +19,14 @@
 from sklearn.metrics import adjusted_rand_score
 from sklearn.cluster import SpectralClustering
 from Space.cons_func import get_results, get_domains
-from Space.utils import calculate_location_adj, plot_results_ari, get_bool_martix
-slide_id = "2"
+from Space.utils import calculate_location_adj, plot_results_ari, get_bool_martix,plot_ari_with_removal
+slide_id = "3"
 adata = sc.read(f"./Data/BARISTASeq/BARISTASeq_Sun2021Integrating_Slice_{slide_id}_data.h5ad")
 adata.var_names_make_unique()
 
 gt = adata.obs["layer"]
 k = 6
-epochs = 300
+epochs = 200
 seed = 666
 alpha = 1
 learning_rate = 0.001
@@ -53,15 +53,21 @@
 # mul_reults = mul_reults.iloc[:, 2:].dropna()
 
 
-collected_results = get_results(func_dict, use_multithreading=False, monitor_performance=True)
-mul_reults = get_domains(adata,collected_results,gt)
-mul_reults= mul_reults.loc[mul_reults["ground_truth"] != 'outside_VISp']
+# collected_results = get_results(func_dict, use_multithreading=False, monitor_performance=True)
+# mul_reults = get_domains(adata,collected_results,gt)
+# mul_reults= mul_reults.loc[mul_reults["ground_truth"] != 'outside_VISp']
+# mul_reults = mul_reults.dropna()
+# mul_reults = mul_reults.drop('ground_truth', axis=1)
+
+mul_reults = pd.read_csv(f"./Data/BARISTASeq/result{slide_id}.csv", header=0, index_col=0)
+mul_reults= mul_reults.loc[mul_reults["Ground Truth"] != 'outside_VISp']
 mul_reults = mul_reults.dropna()
-mul_reults = mul_reults.drop('ground_truth', axis=1)
+mul_reults = mul_reults.iloc[:, 2:]
 
+mul_reults = plot_ari_with_removal(mul_reults,4)
 
 
-plot_results_ari(mul_reults)
+# plot_results_ari(mul_reults)
 outside_VISp_indices = adata.obs["layer"] == "outside_VISp"
 adata = adata[~outside_VISp_indices]
 gt = adata.obs["layer"]