- Support both min famdb *and* current dfam 3.8 famdb

Dfam-consortium · Sep 10, 2024 · 295ed54 · 295ed54
1 parent 00a8318
commit 295ed54
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 4 deletions.
diff --git a/famdb.py b/famdb.py
@@ -538,6 +538,10 @@ def command_append(args):
     """
 
     lookup = args.db_dir.get_all_taxa_names()
+    repbase_lookup = {}
+    with open(REPBASE_FILE) as file:
+        repbase_lookup = json.load(file)
+    lookup.update(repbase_lookup)
 
     header = None
 

diff --git a/famdb_helper_classes.py b/famdb_helper_classes.py
@@ -632,13 +632,17 @@ def set_family_code(family, code, value):
                     for spec in matches.group(1).split(","):
                         name = spec.strip()
                         if name:
-                            name = name.replace("[", "")
-                            name = name.replace("]", "")
-                            tax_id = lookup.get(name.lower())
+                            tax_id = lookup.get(name)
                             if tax_id is not None:
                                 family.clades += [tax_id]
                             else:
-                                LOGGER.warning("Could not find taxon for '%s'", name)
+                                name = name.replace("[", "")
+                                name = name.replace("]", "")
+                                tax_id = lookup.get(name.lower())
+                                if tax_id is not None:
+                                    family.clades += [tax_id]
+                                else:
+                                    LOGGER.warning("Could not find taxon for '%s' upper or lower: line=%s, and ID=%s", name, value, family.accession)
                 matches = re.search(r"SearchStages:\s*(\S+)", value)
                 if matches:
                     family.search_stages = matches.group(1).strip()