Merge pull request #72 from gustaveroussy/dev

CHANGELOG.md

@@ -1,3 +1,10 @@
+## [1.0.14] - 2024-04-22
+
+### Changed
+- Renamed `embed_wsi_patches` to `infer_wsi_patches`
+- `infer_wsi_patches` now accepts also callables
+- Improves tile gathering speed and decreases overall computation of tile-wise inference
+
 ## [1.0.13] - 2024-04-22
 
 ### Changed

docs/api/patches.md

@@ -1,5 +1,5 @@
 ::: sopa.patches.Patches2D
 
-::: sopa.patches.infer.embed_wsi_patches
+::: sopa.patches.infer.infer_wsi_patches
 
 ::: sopa.patches.cluster.cluster_embeddings

docs/faq.md

@@ -36,7 +36,7 @@ In this documentation, `data_path` denotes the path to your raw data. Select the
     - a `Morphology_ChannelID_Dictionary.txt` file containing channel names
     - a `Morphology2D` directory containing the images, end in `_F*.TIF`.
 
-    These files must be exported as flat files in AtomX. In brief, you should have this file structure:
+    These files must be exported as flat files in AtomX. That is: within a study, click on "Export" and then select files from the "Flat CSV Files" section (transcripts flat and FOV position flat). You should have this file structure:
 
     ```txt
     .

docs/tutorials/he.ipynb

@@ -170,7 +170,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "from sopa.patches.infer import embed_wsi_patches\n",
+    "from sopa.patches.infer import infer_wsi_patches\n",
     "from sopa.patches.cluster import cluster_embeddings"
    ]
   },
@@ -191,7 +191,7 @@
     }
    ],
    "source": [
-    "embedding = embed_wsi_patches(\n",
+    "embedding = infer_wsi_patches(\n",
     "    sdata, \"Resnet50Features\", patch_width=250, level=1, image_key=he_image_key\n",
     ")"
    ]
@@ -200,7 +200,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "Then, clustering can be run on the patches embeddings. This will generate a `GeoDataFrame` under `sdata[\"sopa_embedding_patches\"]`, as below. Note the `\"cluster\"` column that has been added."
+    "Then, clustering can be run on the patches embeddings. This will generate a `GeoDataFrame` under `sdata[\"sopa_patches_inference\"]`, as below. Note the `\"cluster\"` column that has been added."
    ]
   },
   {
@@ -364,7 +364,7 @@
    "source": [
     "cluster_embeddings(sdata, 'sopa_Resnet50Features')\n",
     "\n",
-    "sdata[\"sopa_embedding_patches\"]"
+    "sdata[\"sopa_patches_inference\"]"
    ]
   },
   {
@@ -403,7 +403,7 @@
    "source": [
     "sdata\\\n",
     "    .pl.render_shapes(\"region_of_interest\", outline=True, fill_alpha=0, outline_color=\"#333\")\\\n",
-    "    .pl.render_shapes(\"sopa_embedding_patches\", color=\"cluster\")\\\n",
+    "    .pl.render_shapes(\"sopa_patches_inference\", color=\"cluster\")\\\n",
     "    .pl.show(\"global\")"
    ]
   },
@@ -431,7 +431,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "res_gdf = sopa.spatial.sjoin(sdata, \"cell_boundaries\", \"sopa_embedding_patches\", target_coordinate_system=\"global\")"
+    "res_gdf = sopa.spatial.sjoin(sdata, \"cell_boundaries\", \"sopa_patches_inference\", target_coordinate_system=\"global\")"
    ]
   },
   {
@@ -504,7 +504,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.12"
+   "version": "3.6.8"
   }
  },
  "nbformat": 4,

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "sopa"
-version = "1.0.13"
+version = "1.0.14"
 description = "Spatial-omics pipeline and analysis"
 documentation = "https://gustaveroussy.github.io/sopa"
 homepage = "https://gustaveroussy.github.io/sopa"

sopa/_constants.py

@@ -8,7 +8,7 @@ class SopaKeys:
 
     BOUNDS = "bboxes"
     PATCHES_ILOCS = "ilocs"
-    EMBEDDINGS_PATCHES_KEY = "sopa_embedding_patches"
+    PATCHES_INFERENCE_KEY = "sopa_patches_inference"
 
     UNS_KEY = "sopa_attrs"
     UNS_HAS_TRANSCRIPTS = "transcripts"

sopa/patches/cluster.py

@@ -53,7 +53,7 @@ def cluster_embeddings(
         ), f"Method {method} is not available. Use one of: {', '.join(METHODS_DICT.keys())}"
         method = METHODS_DICT[method]
 
-    gdf_patches = sdata[SopaKeys.EMBEDDINGS_PATCHES_KEY]
+    gdf_patches = sdata[SopaKeys.PATCHES_INFERENCE_KEY]
 
     ilocs = np.array(list(gdf_patches.ilocs))
     embeddings = element.compute().data[:, ilocs[:, 1], ilocs[:, 0]].T