fix(design): allow additional file extensions in GEOFiles and improve…

… processed data files handling

seqnado/design.py

@@ -929,6 +929,7 @@ class GEOFiles(BaseModel):
     sample_names: List[str]
     config: dict
     design: pd.DataFrame
+    extensions_allowed: List[str] = [".txt", ".bigWig", ".bed", ".tsv", ".vcf.gz"]
 
     processed_files: Optional[List[Union[str, pathlib.Path]]] = None
 
@@ -947,7 +948,8 @@ def md5sums(self):
 
     @property
     def processed_data_files(self) -> pd.DataFrame:
-        wanted_exts = [".txt", ".bigWig", ".bed", '.tsv']
+
+        wanted_exts = self.extensions_allowed
         unwanted_files = [*self.md5sums]
 
         # Create a DataFrame with the processed files

seqnado/workflow/rules/geo_submission.smk

@@ -37,8 +37,12 @@ def get_symlinked_files(wc: Any = None) -> List[str]:
                          processed_files=[str(p) for p in OUTPUT.files],
                          )
 
-    processed_files = geo_files.processed_data_files['output_file_name'].tolist()
-    processed_files = [outdir / fn for fn in processed_files]
+
+
+    if processed_files := geo_files.processed_data_files['output_file_name'].tolist():
+        processed_files = [outdir / fn for fn in processed_files]
+    else:
+        processed_files = []
 
 
     return [*fastqs, *processed_files]