fix smk pipeline after cli update

Author: quentinblampey

sopa/cli/app.py

@@ -175,7 +175,8 @@ def aggregate(
         help="Optional image key of the SpatialData object. By default, considers the only one image. It can be useful if another image is added later on",
     ),
     method_name: str = typer.Option(
-        help="If segmentation was performed with a generic method, this is the name of the method used."
+        None,
+        help="If segmentation was performed with a generic method, this is the name of the method used.",
     ),
 ):
     """Create an `anndata` table containing the transcript count and/or the channel intensities per cell"""

workflow/Snakefile

@@ -56,7 +56,7 @@ checkpoint patchify_baysor:
         paths.smk_cellpose_boundaries if args.cellpose else [],
     output:
         patches_file = paths.smk_patches_file_baysor,
-        baysor_temp_dir = directory(paths.baysor_temp_dir),
+        smk_baysor_temp_dir = directory(paths.smk_baysor_temp_dir),
     params:
         args_patchify = str(args["patchify"].where(contains="micron")),
         args_baysor = args.dump_baysor_patchify() if args.baysor else "",
@@ -73,23 +73,23 @@ rule patch_segmentation_cellpose:
         paths.smk_patches_file_image,
         paths.smk_patches,
     output:
-        f"{paths.cellpose_temp_dir}/{{index}}.parquet",
+        f"{paths.smk_cellpose_temp_dir}/{{index}}.parquet",
     conda:
         "sopa"
     params:
         args_cellpose = str(args["segmentation"]["cellpose"]),
     shell:
         """
-        sopa segmentation cellpose {paths.sdata_path} --patch-dir {paths.cellpose_temp_dir} --patch-index {wildcards.index} {params.args_cellpose}
+        sopa segmentation cellpose {paths.sdata_path} --patch-dir {paths.smk_cellpose_temp_dir} --patch-index {wildcards.index} {params.args_cellpose}
         """
 
 rule patch_segmentation_baysor:
     input:
         patches_file = paths.smk_patches_file_baysor,
-        baysor_patch = f"{paths.baysor_temp_dir}/{{index}}",
+        baysor_patch = f"{paths.smk_baysor_temp_dir}/{{index}}",
     output:
-        f"{paths.baysor_temp_dir}/{{index}}/segmentation_polygons.json",
-        f"{paths.baysor_temp_dir}/{{index}}/segmentation_counts.loom",
+        f"{paths.smk_baysor_temp_dir}/{{index}}/segmentation_polygons.json",
+        f"{paths.smk_baysor_temp_dir}/{{index}}/segmentation_counts.loom",
     params:
         args_baysor_prior_seg = args.baysor_prior_seg,
     resources:
@@ -119,7 +119,7 @@ rule resolve_cellpose:
         "sopa"
     shell:
         """
-        sopa resolve cellpose {paths.sdata_path} --patch-dir {paths.cellpose_temp_dir}
+        sopa resolve cellpose {paths.sdata_path} --patch-dir {paths.smk_cellpose_temp_dir}
         """
 
 rule resolve_baysor:
@@ -138,7 +138,7 @@ rule resolve_baysor:
         """
         sopa resolve baysor {paths.sdata_path} --gene-column {args.gene_column} {params.args_patches_dirs} {params.args_shapes}
 
-        rm -r {paths.baysor_temp_dir}    # cleanup large baysor files
+        rm -r {paths.smk_baysor_temp_dir}    # cleanup large baysor files
         """
 
 rule aggregate:

workflow/utils.py

@@ -26,55 +26,62 @@ class WorkflowPaths:
     def __init__(self, config: dict) -> None:
         self.config = _sanity_check_config(config)
 
+        # spatialdata object files
         self.sdata_path = Path(self.config["sdata_path"])
         self.data_path = self.config["data_path"]
         self.sdata_zgroup = self.sdata_path / ".zgroup"  # trick to fix snakemake ChildIOException
 
+        # spatial elements directories
         self.shapes_dir = self.sdata_path / "shapes"
         self.points_dir = self.sdata_path / "points"
         self.images_dir = self.sdata_path / "images"
         self.table_dir = self.sdata_path / "table"
 
+        # sdata.shapes
         self.baysor_boundaries = self.shapes_dir / "baysor_boundaries"
         self.cellpose_boundaries = self.shapes_dir / "cellpose_boundaries"
         self.patches = self.shapes_dir / "sopa_patches"
 
+        # snakemake cache / intermediate files
         self.sopa_cache = self.sdata_path / ".sopa_cache"
         self.smk_table = self.sopa_cache / "table"
         self.smk_patches = self.sopa_cache / "patches"
         self.smk_patches_file_image = self.sopa_cache / "patches_file_image"
         self.smk_patches_file_baysor = self.sopa_cache / "patches_file_baysor"
-        self.smk_cellpose_boundaries = self.sopa_cache / "cellpose_boundaries"
-        self.smk_baysor_boundaries = self.sopa_cache / "baysor_boundaries"
+        self.smk_cellpose_temp_dir = self.sopa_cache / "cellpose_boundaries"
+        self.smk_baysor_temp_dir = self.sopa_cache / "baysor_boundaries"
+        self.smk_cellpose_boundaries = self.sopa_cache / "cellpose_boundaries_done"
+        self.smk_baysor_boundaries = self.sopa_cache / "baysor_boundaries_done"
         self.smk_aggregation = self.sopa_cache / "aggregation"
 
+        # annotation files
         self.annotations = []
         if "annotation" in self.config:
             key = self.config["annotation"].get("args", {}).get("cell_type_key", "cell_type")
             self.annotations = self.table_dir / "table" / "obs" / key
 
-        self.cellpose_temp_dir = self.sopa_cache / "cellpose_boundaries"
-        self.baysor_temp_dir = self.sopa_cache / "baysor_boundaries"
-
+        # user-friendly output files
         self.explorer_directory = self.sdata_path.with_suffix(".explorer")
         self.explorer_directory.mkdir(parents=True, exist_ok=True)
-
         self.explorer_experiment = self.explorer_directory / "experiment.xenium"
         self.explorer_image = self.explorer_directory / "morphology.ome.tif"
-
         self.report = self.explorer_directory / "analysis_summary.html"
 
     def cells_paths(self, file_content: str, name, dirs: bool = False):
         if name == "cellpose":
-            return [str(self.cellpose_temp_dir / f"{i}.parquet") for i in range(int(file_content))]
+            return [
+                str(self.smk_cellpose_temp_dir / f"{i}.parquet") for i in range(int(file_content))
+            ]
         if name == "baysor":
             indices = map(int, file_content.split())
             BAYSOR_FILES = ["segmentation_polygons.json", "segmentation_counts.loom"]
 
             if dirs:
-                return [str(self.baysor_temp_dir / str(i)) for i in indices]
+                return [str(self.smk_baysor_temp_dir / str(i)) for i in indices]
             return [
-                str(self.baysor_temp_dir / str(i) / file) for i in indices for file in BAYSOR_FILES
+                str(self.smk_baysor_temp_dir / str(i) / file)
+                for i in indices
+                for file in BAYSOR_FILES
             ]
 
 
@@ -114,7 +121,8 @@ def where(self, keys: Optional[list[str]] = None, contains: Optional[str] = None
 
     def dump_baysor_patchify(self):
         return (
-            str(self["segmentation"]["baysor"]) + f" --baysor-temp-dir {self.paths.baysor_temp_dir}"
+            str(self["segmentation"]["baysor"])
+            + f" --baysor-temp-dir {self.paths.smk_baysor_temp_dir}"
         )
 
     @classmethod