handle omics-level failures and missing metatranascriptome

nmdc_automation/re_iding/base.py

@@ -381,11 +381,15 @@ def update_metatranscriptome_activity_set(self, db_record: Dict,
         Return a new Database instance with the metatranscriptome_activity_set
         and its data objects updated to new IDs.
         """
+        metatranscriptome_records = db_record.get(METATRANSCRIPTOME_ACTIVITY_SET, [])
+        if not metatranscriptome_records:
+            logger.info(f"No metatranscriptome_activity_set found for {db_record[OMICS_PROCESSING_SET][0]['id']}")
+            return new_db
         logger.info(f"Updating metatranscriptome_activity_set for "
                     f"{db_record[OMICS_PROCESSING_SET][0]['id']}")
         new_omics_processing = new_db.omics_processing_set[0]
 
-        for metatranscriptome_rec in db_record.get(METATRANSCRIPTOME_ACTIVITY_SET, []):
+        for metatranscriptome_rec in metatranscriptome_records:
             # old records have non-conforming type e.g. nmdc:MetaT,
             # nmdc:metaT etc. - fix it
             activity_type = "nmdc:MetatranscriptomeActivity"

nmdc_automation/re_iding/scripts/re_id_tool.py

@@ -260,9 +260,12 @@ def extract_records(ctx, study_id, api_base_url):
             if passing_records:
                 db.__setattr__(set_name, passing_records)
 
-        retrieved_databases.append(db)
+
+
         if is_failed_data:
             retrieved_failed_databases.append(db_failed)
+        else:
+            retrieved_databases.append(db)
 
     json_data = json.loads(json_dumper.dumps(retrieved_databases, inject_type=False))
     db_outfile = DATA_DIR.joinpath(f"{study_id}_associated_record_dump.json")