Add cosmx liver dataset (#66)

Author: LouisK92

scripts/create_resources/spatial/process_bruker_cosmx.sh

@@ -23,6 +23,16 @@ param_list:
     dataset_organism: "mus_musculus"
     segmentation_id: ["cell"]
 
+  - id: "bruker_cosmx/bruker_human_liver_cosmx"
+    input_raw: "https://smi-public.objects.liquidweb.services/NormalLiverFiles.zip"
+    input_flat_files: "https://syncandshare.desy.de/index.php/s/zYT4fM28y86cZeW/download/NormalLiver.zip"
+    dataset_name: "Bruker CosMx Human Liver"
+    dataset_url: "https://nanostring.com/products/cosmx-spatial-molecular-imager/ffpe-dataset/human-liver-rna-ffpe-dataset/"
+    dataset_summary: "Bruker CosMx Human Liver dataset on FFPE."
+    dataset_description: "Bruker CosMx Human Liver dataset on FFPE."
+    dataset_organism: "homo_sapiens"
+    segmentation_id: ["cell"]
+
 
 output_dataset: "\$id/dataset.zarr"
 output_state: "\$id/state.yaml"

src/datasets/loaders/bruker_cosmx/script.py

@@ -142,13 +142,13 @@
 else:
     print(datetime.now() - t0, "Download and extract flat files", flush=True)
     os.system(f"wget {par['input_flat_files']} -O '{FILE_NAME_FLAT}'")
-
+    
     with zipfile.ZipFile(FILE_NAME_FLAT, 'r') as zip_ref:
         zip_ref.extractall(TMP_DIR)
-
+    
     print(datetime.now() - t0, f"Move flat files to {DATA_DIR}", flush=True)
     source_dir = FILE_NAME_FLAT.parent / FILE_NAME_FLAT.stem
-
+    
     file_names = os.listdir(source_dir)
     for file_name in file_names:
         if not (DATA_DIR / file_name).exists():
@@ -164,9 +164,9 @@
     print(datetime.now() - t0, "Create CellLabels folder with CellLabels tif", flush=True)
     # Create CellLabels folder with CellLabels tif (somehow this folder name is expected and this is not always present)
     # see e.g. late discussion in https://github.com/gustaveroussy/sopa/issues/285
-
+    
     labels_dir.mkdir(parents=True, exist_ok=True)
-
+    
     # Get all folders in data_dir that start with "FOV" and move the CellLabels_FXXX.tif file to the CellLabels folder
     print(datetime.now() - t0, "Move CellLabels_FXXX.tif files to CellLabels folder", flush=True)
     for fov_dir in DATA_DIR.glob("FOV*"):