Fix/stardist

src/base/labels_nebius.config

@@ -124,8 +124,8 @@ withLabel: veryhightime { time = 24.h }
 
   // similarity metric does not need veryhighmem resources
   withName: '.*similarity_process' {
-    memory = '50.GB'
-    disk = '50.GB'
+    memory = '100.GB'
+    disk = '100.GB'
   }
 }
 

src/methods_segmentation/custom_segmentation/script.py

@@ -2,6 +2,7 @@
 import anndata as ad
 import os
 import shutil
+import pandas as pd
 
 ## VIASH START
 par = {
@@ -20,14 +21,18 @@
 assert par["labels_key"] in sdata.labels, f"Key '{par['labels_key']}' not found in input data."
 
 print(f"Copy segmentation from '{par['labels_key']}'", flush=True)
+metadata = sdata.tables["metadata"]
+# Select only the columns that exist — Xenium provides cell_id and region,
+# Vizgen uses different column names (or an empty obs) so we take what's available.
+obs_cols = [c for c in ["cell_id", "region"] if c in metadata.obs.columns]
 sdata_segmentation_only = sd.SpatialData(
   labels={
     "segmentation": sdata[par["labels_key"]]
   },
   tables={
     "table": ad.AnnData(
-      obs=sdata.tables["metadata"].obs[["cell_id", "region"]],
-      var=sdata.tables["metadata"].var[[]]
+      obs=metadata.obs[obs_cols],
+      var=metadata.var[[]]
     )
   }
 )

src/methods_segmentation/stardist/config.vsh.yaml

@@ -34,6 +34,7 @@ engines:
           - stardist
           - tensorflow==2.17.0
           - numpy<2.0.0
+          - scipy<1.15.0
   - type: native
 
 runners:

src/methods_transcript_assignment/basic_transcript_assignment/script.py

@@ -1,6 +1,7 @@
 import numpy as np
 import xarray as xr
 import dask
+import dask.dataframe as dd
 import spatialdata as sd
 import anndata as ad
 import pandas as pd
@@ -33,14 +34,14 @@
 assert par['coordinate_system'] in segmentation_coord_systems, f"Coordinate system '{par['coordinate_system']}' not found in input data."
 
 print('Transforming transcripts coordinates', flush=True)
-# Parquet partitions each start from index 0, causing duplicate index values in the
-# combined dask DataFrame. sd.transform() internally builds pd.Series(..., index=transformed.index)
-# which fails with "cannot reindex on an axis with duplicate labels".
-# Fix: reset to a global monotonic index before transforming; restore attrs explicitly
-# because reset_index() drops them, which would break spatialdata's PointsModel check.
+# Multi-partition parquet files each start with a 0-based index, producing duplicate index
+# values in the combined dask DataFrame. sd.transform() internally creates a pd.Series with
+# index=transformed.index; when that dask index is computed it triggers an assign expression
+# that fails on duplicate/lazy indices. Fix: materialize to pandas and rebuild as a single
+# dask partition with a clean RangeIndex before transforming.
 # The original sdata[transcripts_key] is left unchanged so lines below remain consistent.
 transcripts_input = sdata[par['transcripts_key']]
-transcripts_reset = transcripts_input.reset_index(drop=True)
+transcripts_reset = dd.from_pandas(transcripts_input.compute().reset_index(drop=True), npartitions=1)
 transcripts_reset.attrs.update(transcripts_input.attrs)
 transcripts = sd.transform(transcripts_reset, to_coordinate_system=par['coordinate_system'])