Dev

sopa/annotation/tangram/run.py

@@ -14,7 +14,7 @@
     import tangram as tg
 except ImportError:
     raise ImportError(
-        "To use tangram, you need its corresponding sopa extra: `pip install 'sopa[tangram]'`"
+        "To use tangram, you need its corresponding sopa extra: `pip install 'sopa[tangram]'` (normal mode) or `pip install -e '.[tangram]'` (if using snakemake)"
     )
 
 from sopa._constants import SopaKeys

sopa/embedding/patches.py

@@ -6,7 +6,7 @@
     import torch
 except ImportError:
     raise ImportError(
-        "For patch embedding, you need `torch` (and perhaps `torchvision`). Consider installing the sopa WSI extra: `pip install 'sopa[wsi]'`"
+        "For patch embedding, you need `torch` (and perhaps `torchvision`). Consider installing the sopa WSI extra: `pip install 'sopa[wsi]'` (normal mode) or `pip install -e '.[wsi]'` (if using snakemake)"
     )
 
 import numpy as np

sopa/io/reader/cosmx.py

@@ -58,7 +58,7 @@ def cosmx(
     if read_proteins:
         protein_dir_dict = {
             int(protein_dir.parent.name[3:]): protein_dir
-            for protein_dir in list(path.rglob("**/ProteinImages"))
+            for protein_dir in list(path.rglob("**/FOV*/ProteinImages"))
         }
         assert len(protein_dir_dict), f"No directory called 'ProteinImages' was found under {path}"
 

sopa/io/reader/wsi.py

@@ -49,12 +49,15 @@ def wsi(
 
     multiscale_image = MultiscaleSpatialImage.from_dict(images)
     multiscale_image.attrs["metadata"] = tiff_metadata
-    multiscale_image.name = image_name
 
     if as_image:
+        multiscale_image.name = image_name
         return multiscale_image
 
-    return SpatialData(images={image_name: multiscale_image})
+    sdata = SpatialData(images={image_name: multiscale_image})
+    sdata[image_name].attrs["metadata"] = tiff_metadata
+
+    return sdata
 
 
 def _get_scale_transformation(scale_factor: float):

sopa/segmentation/baysor/prepare.py

@@ -4,7 +4,7 @@
     import toml
 except ImportError:
     raise ImportError(
-        "To use baysor, you need its corresponding sopa extra: `pip install 'sopa[baysor]'`.\
+        "To use baysor, you need its corresponding sopa extra: `pip install 'sopa[baysor]'` (normal mode) or `pip install -e '.[baysor]'` (if using snakemake).\
         \nAlso, make sure to install the baysor executable (https://github.com/kharchenkolab/Baysor)."
     )
 

sopa/segmentation/baysor/resolve.py

@@ -133,7 +133,7 @@ def resolve(
 
     image_key = get_key(sdata, "images")
     points = get_element(sdata, "points")
-    transformations = get_transformation(points, get_all=True)
+    transformations = get_transformation(points, get_all=True).copy()
 
     geo_df = ShapesModel.parse(geo_df, transformations=transformations)
 

sopa/segmentation/methods.py

@@ -30,7 +30,7 @@ def cellpose_patch(
         from cellpose import models
     except ImportError:
         raise ImportError(
-            "To use cellpose, you need its corresponding sopa extra: `pip install 'sopa[cellpose]'`"
+            "To use cellpose, you need its corresponding sopa extra: `pip install 'sopa[cellpose]'` (normal mode) or `pip install -e '.[cellpose]'` (if using snakemake)"
         )
 
     cellpose_model_kwargs = cellpose_model_kwargs or {}

sopa/segmentation/patching.py

@@ -207,7 +207,7 @@ def write(self, overwrite: bool = True, shapes_key: str | None = None) -> gpd.Ge
             }
         )
         geo_df = ShapesModel.parse(
-            geo_df, transformations=get_transformation(self.element, get_all=True)
+            geo_df, transformations=get_transformation(self.element, get_all=True).copy()
         )
 
         self.sdata.shapes[shapes_key] = geo_df

sopa/segmentation/stainings.py

@@ -177,7 +177,9 @@ def add_shapes(cls, sdata: SpatialData, cells: list[Polygon], image_key: str, sh
         geo_df = gpd.GeoDataFrame({"geometry": cells})
         geo_df.index = image_key + geo_df.index.astype(str)
 
-        geo_df = ShapesModel.parse(geo_df, transformations=get_transformation(image, get_all=True))
+        geo_df = ShapesModel.parse(
+            geo_df, transformations=get_transformation(image, get_all=True).copy()
+        )
         sdata.shapes[shapes_key] = geo_df
         save_shapes(sdata, shapes_key, overwrite=True)
 

sopa/utils/polygon_crop.py

@@ -132,7 +132,7 @@ def polygon_selection(
     geo_df = gpd.GeoDataFrame(geometry=[polygon])
 
     geo_df = ShapesModel.parse(
-        geo_df, transformations=get_transformation(sdata[image_key], get_all=True)
+        geo_df, transformations=get_transformation(sdata[image_key], get_all=True).copy()
     )
     sdata.shapes[ROI.KEY] = geo_df
     save_shapes(sdata, ROI.KEY)

workflow/config/cosmx/cellpose_baysor.yaml

@@ -34,7 +34,7 @@ segmentation:
       segmentation:
         scale: 60 # typical cell radius
         scale_std: "25%" # cell radius standard deviation
-        prior_segmentation_confidence: 0
+        prior_segmentation_confidence: 1
         estimate_scale_from_centers: false
         n_clusters: 4
         iters: 500

workflow/config/merscope/baysor_cellpose.yaml

@@ -35,7 +35,7 @@ segmentation:
       segmentation:
         scale: 6.25 # typical cell radius
         scale_std: "25%" # cell radius standard deviation
-        prior_segmentation_confidence: 0.75 # confidence of the cellpose confidence (float in [0, 1])
+        prior_segmentation_confidence: 1 # confidence of the cellpose confidence (float in [0, 1])
         estimate_scale_from_centers: false
         n_clusters: 4
         iters: 500

workflow/config/xenium/cellpose_baysor.yaml

@@ -34,7 +34,7 @@ segmentation:
       segmentation:
         scale: 6.25 # typical cell radius
         scale_std: "25%" # cell radius standard deviation
-        prior_segmentation_confidence: 0.75 # confidence of the cellpose confidence (float in [0, 1])
+        prior_segmentation_confidence: 1 # confidence of the cellpose confidence (float in [0, 1])
         estimate_scale_from_centers: false
         n_clusters: 4
         iters: 500