corrected errors around novae execution and tangram map training data row indices

Author: ctastad

sopa/utils/annotation.py

@@ -343,7 +343,8 @@ def run_group(self, level: int = 0, indices_sp=None, indices_sc=None, output_dir
                     # Save as CSV
                     from pathlib import Path
                     output_path = Path(output_dir) / f"split_{i}_training_data.csv"
-                    train_df.to_csv(output_path, index=False)
+                    train_df.index.name = 'gene_name'
+                    train_df.to_csv(output_path, index=True)
                     log.info(f"Saved training data to {output_path}")
                 else:
                     log.warning("No training data found in ad_map.uns['train_genes_df']")

workflow/Snakefile

@@ -288,7 +288,7 @@ rule novae:
         model = config["novae"]["model"],
         zero_shot = "True" if config["novae"]["zero_shot"] else "False",
         num_workers = config["novae"]["num_workers"],
-        script = "workflow/src/run_novae.py",
+        script = "src/run_novae.py",
     shell:
         """
         python {params.script} {params.sdata_path} {params.model} {params.radius} {params.zero_shot} {params.num_workers}