Fixing Scanpy 1.9.3 error for mixed datatype column

scanpy-scripts-tests.bats

@@ -180,7 +180,7 @@ setup() {
         skip "$singlet_obs exists"
     fi
 
-    run rm -rf $batch_obs && echo -e "batch\n$(printf "%0.sbatch1\n" {1..1350})\n$(printf "%0.sbatch2\n" {1..1350})" > $batch_obs
+    run rm -rf $batch_obs && echo -e "batch\tadditional_column\n$(for i in {1..1350}; do echo -e "batch1\tdata$i"; done)\n$(for i in {1..1350}; do echo -e "batch2\tinfo$i"; done)" > $batch_obs
 
     [ "$status" -eq 0 ]
     [ -f "$batch_obs" ]

scanpy_scripts/lib/_read.py

@@ -42,4 +42,26 @@ def read_10x(
             right_index=True,
             suffixes=(False, False),
         )
+
+    # Convert mixed dtype columns to 'string' type to preserve all information
+    obs_mixed_columns = columns_with_multiple_dtypes(adata.obs)
+
+    for column in obs_mixed_columns:
+        adata.obs[column] = adata.obs[column].astype("str")
+
+    var_mixed_columns = columns_with_multiple_dtypes(adata.var)
+
+    for column in var_mixed_columns:
+        adata.var[column] = adata.var[column].astype("str")
+
     return adata
+
+
+def columns_with_multiple_dtypes(df):
+    mixed_dtype_columns = []
+    for column in df.columns:
+        # Get unique dtypes in the column
+        unique_dtypes = df[column].apply(type).unique()
+        if len(unique_dtypes) > 1:
+            mixed_dtype_columns.append(column)
+    return mixed_dtype_columns