fix annotation and adjust logging level of a few messages

Author: rpetit3

bactopia/cli/summary.py

@@ -404,14 +404,14 @@ def summary(
                         dfs.append(df)
                         logging.debug(f"\tRank: {rank} ({reason})")
                     else:
-                        logging.debug(
+                        logging.info(
                             f"Skipping {sample['id']} ({sample['path']}) due to missing files. Missing:"
                         )
                         for missing_file in parsable_files:
-                            logging.debug(f"\t{missing_file}")
+                            logging.info(f"\t{missing_file}")
                         increment_and_append("ignore-unknown", sample["id"])
             else:
-                logging.debug(
+                logging.info(
                     f"Skipping {sample['id']} ({sample['path']}), incomplete or not a Bactopia directory"
                 )
                 increment_and_append("ignore-unknown", sample["id"])

bactopia/parsers/annotator.py

@@ -57,7 +57,8 @@ def _parse_annotation(path: str, name: str) -> dict:
     with open(path, "rt") as fh:
         for line in fh:
             line = line.rstrip()
-            key, val = line.split(":")
-            if key in COLS:
-                results[f"annotator_total_{key}"] = int(val.lstrip())
+            if ":" in line:
+                key, val = line.split(":")
+                if key in COLS:
+                    results[f"annotator_total_{key}"] = int(val.lstrip())
     return results

bactopia/parsers/parsables.py

@@ -1,6 +1,7 @@
 """
 A list of files that can be parsed by Bactopia
 """
+import logging
 from pathlib import Path
 
 EXCLUDE_COLUMNS = [
@@ -24,8 +25,6 @@
 def get_parsable_files(path: str, name: str) -> list:
     parsable_files = {
         # main
-        # annotator
-        f"{path}/main/annotator/prokka/{name}.txt": "annotator",
         # assembler
         f"{path}/main/assembler/{name}.tsv": "assembler",
         # gather
@@ -49,14 +48,23 @@ def get_parsable_files(path: str, name: str) -> list:
 
     # Check annotation files seperately, since Prokka or Bakta can be used
     if Path(f"{path}/main/annotator/bakta/{name}.txt").exists():
+        logging.debug(
+            f"Found Bakta annotation file: {path}/main/annotator/bakta/{name}.txt"
+        )
         parsable_files[f"{path}/main/annotator/bakta/{name}.txt"] = "annotator"
     elif Path(f"{path}/main/annotator/prokka/{name}.txt").exists():
+        logging.debug(
+            f"Found Prokka annotation file: {path}/main/annotator/prokka/{name}.txt"
+        )
         parsable_files[f"{path}/main/annotator/prokka/{name}.txt"] = "annotator"
     else:
         is_complete = False
         missing_files.append(f"{path}/main/annotator/prokka/{name}.txt")
         missing_files.append(f"{path}/main/annotator/bakta/{name}.txt")
 
+    logging.debug(f"Missing Files: {missing_files}")
+    logging.debug(f"Is Complete: {is_complete}")
+
     if is_complete:
         parsable_files[f"{path}/main/qc/summary/{name}-original.json"] = "qc"
         parsable_files[f"{path}/main/qc/summary/{name}-final.json"] = "qc"