only keep NCBIGene or UniProtKB subjects/objects from biogrid

monarch-initiative · Jan 2, 2024 · cfee8c4 · cfee8c4
1 parent 3d1b0a8
commit cfee8c4
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 12 deletions.
diff --git a/src/monarch_ingest/ingests/biogrid/biogrid.py b/src/monarch_ingest/ingests/biogrid/biogrid.py
@@ -14,15 +14,18 @@
 
     publications = get_publication_ids(row['Publication Identifiers'])
 
-    association = PairwiseGeneToGeneInteraction(
-        id="uuid:" + str(uuid.uuid1()),
-        subject=gid_a,
-        predicate="biolink:interacts_with",
-        object=gid_b,
-        has_evidence=evidence,
-        publications=publications,
-        primary_knowledge_source="infores:biogrid",
-        aggregator_knowledge_source=["infores:monarchinitiative"]
-    )
+    # Only keep interactions using NCBIGene or UniProtKB identifiers, could also filter on taxid
+    if gid_a.startswith("NCBIGene:") or gid_a.startswith("UniProtKB:") \
+            and gid_b.startswith("NCBIGene:") or gid_b.startswith("UniProtKB:"):
+        association = PairwiseGeneToGeneInteraction(
+            id="uuid:" + str(uuid.uuid1()),
+            subject=gid_a,
+            predicate="biolink:interacts_with",
+            object=gid_b,
+            has_evidence=evidence,
+            publications=publications,
+            primary_knowledge_source="infores:biogrid",
+            aggregator_knowledge_source=["infores:monarchinitiative"]
+        )
 
-    koza_app.write(association)
+        koza_app.write(association)
diff --git a/src/monarch_ingest/ingests/biogrid/biogrid_util.py b/src/monarch_ingest/ingests/biogrid/biogrid_util.py
@@ -14,7 +14,10 @@ def get_gene_id(raw_id: str) -> str:
     :param raw_id: str, raw BioGRID input string (a pseudo-CURIE)
     :return:
     """
-    gid = raw_id.replace("entrez gene/locuslink", "NCBIGene")
+    gid = (raw_id
+           .replace("entrez gene/locuslink:", "NCBIGene:")
+           .replace("uniprot/swiss-prot:", "UniProtKB:"))
+
     return gid