From cb627eca6fd3c06f38dcec1e45489164552a1129 Mon Sep 17 00:00:00 2001
From: Kevin Schaper <kevinschaper@gmail.com>
Date: Wed, 11 Dec 2024 18:45:19 -0800
Subject: [PATCH] Use Xenbase GenePageToGene mapping to create orthology
 records for each species of xenopus rather than the dangling genepage
 orthology records we were making

---
 .../ingests/xenbase/orthologs.py              | 37 +++++++++++--------
 .../ingests/xenbase/orthologs.yaml            |  3 ++
 src/monarch_ingest/qc_expect.yaml             |  2 +
 tests/unit/xenbase/test_xenbase_orthologs.py  | 26 ++++++++++---
 4 files changed, 47 insertions(+), 21 deletions(-)

diff --git a/src/monarch_ingest/ingests/xenbase/orthologs.py b/src/monarch_ingest/ingests/xenbase/orthologs.py
index 609a27ef..5ca4b9da 100644
--- a/src/monarch_ingest/ingests/xenbase/orthologs.py
+++ b/src/monarch_ingest/ingests/xenbase/orthologs.py
@@ -12,31 +12,36 @@
 
 koza_app = get_koza_app("xenbase_orthologs")
 
+genepage_to_gene_map = koza_app.get_map("genepage-2-gene")
+
 while (row := koza_app.get_row()) is not None:
 
     try:
-        gene_id = row['xb_genepage_id']
-        assert gene_id
+        genepage_id = row['xb_genepage_id']
+        assert genepage_id
 
         predicate = "biolink:orthologous_to"
 
         ortholog_id = row['entrez_id']
         assert ortholog_id
 
-        # Instantiate the instance of Gene-to-Gene Homology Association
-        association = GeneToGeneHomologyAssociation(
-            id=f"uuid:{str(uuid.uuid1())}",
-            subject=f"Xenbase:{gene_id}",
-            predicate=predicate,
-            object=f"NCBIGene:{ortholog_id}",
-            aggregator_knowledge_source=["infores:monarchinitiative"],
-            primary_knowledge_source="infores:xenbase",
-            knowledge_level=KnowledgeLevelEnum.knowledge_assertion,
-            agent_type=AgentTypeEnum.manual_agent,
-        )
-
-        # Write the captured Association out
-        koza_app.write(association)
+        gene_ids = genepage_to_gene_map.get(genepage_id).values()
+
+        for gene_id in gene_ids:
+            # Instantiate the instance of Gene-to-Gene Homology Association
+            association = GeneToGeneHomologyAssociation(
+                id=f"uuid:{str(uuid.uuid1())}",
+                subject=f"Xenbase:{gene_id}",
+                predicate=predicate,
+                object=f"NCBIGene:{ortholog_id}",
+                aggregator_knowledge_source=["infores:monarchinitiative"],
+                primary_knowledge_source="infores:xenbase",
+                knowledge_level=KnowledgeLevelEnum.knowledge_assertion,
+                agent_type=AgentTypeEnum.manual_agent,
+            )
+
+            # Write the captured Association out
+            koza_app.write(association)
 
     except (RuntimeError, AssertionError) as rte:
         logger.debug(f"{str(rte)} in data row:\n\t'{str(row)}'")
diff --git a/src/monarch_ingest/ingests/xenbase/orthologs.yaml b/src/monarch_ingest/ingests/xenbase/orthologs.yaml
index ae34277f..6070b341 100644
--- a/src/monarch_ingest/ingests/xenbase/orthologs.yaml
+++ b/src/monarch_ingest/ingests/xenbase/orthologs.yaml
@@ -25,6 +25,9 @@ columns:
  - 'xb_gene_symbol'
  - 'xb_gene_name'
 
+depends_on:
+  - './src/monarch_ingest/maps/genepage-2-gene.yaml'
+  
 edge_properties:
   - 'id'
   - 'category'
diff --git a/src/monarch_ingest/qc_expect.yaml b/src/monarch_ingest/qc_expect.yaml
index 209e58df..6f194e34 100644
--- a/src/monarch_ingest/qc_expect.yaml
+++ b/src/monarch_ingest/qc_expect.yaml
@@ -65,6 +65,8 @@ edges:
       min: 1420000
     xenbase_gene_to_phenotype_edges:
       min: 2000
+    xenbase_orthologs_edges:
+      min: 136746
     alliance_phenotype_edges:
       min: 650000
     alliance_disease_edges:
diff --git a/tests/unit/xenbase/test_xenbase_orthologs.py b/tests/unit/xenbase/test_xenbase_orthologs.py
index 71cecced..e9574576 100644
--- a/tests/unit/xenbase/test_xenbase_orthologs.py
+++ b/tests/unit/xenbase/test_xenbase_orthologs.py
@@ -24,7 +24,19 @@ def script():
 
 
 @pytest.fixture
-def orthology_record(mock_koza, source_name, script, global_table):
+def map_cache():
+    return { 
+        'genepage-2-gene': {
+            'XB-GENEPAGE-478063': { 
+                'tropicalis_id':'XB-GENE-478064',
+                'laevis_l_id': 'XB-GENE-6461998',
+                'laevis_s_id': 'XB-GENE-17342561',
+            }
+        }
+    }
+
+@pytest.fixture
+def orthology_record(mock_koza, source_name, script, global_table, map_cache):
     row = {
         'entrez_id': "8928",
         'xb_genepage_id': "XB-GENEPAGE-478063",
@@ -35,16 +47,20 @@ def orthology_record(mock_koza, source_name, script, global_table):
         name=source_name,
         data=row,
         transform_code=script,
+        map_cache=map_cache,
         global_table=global_table,
     )
 
-
-def test_orthology_record(orthology_record):
+@pytest.mark.parametrize("index, expected_subject", 
+                         [(0, "Xenbase:XB-GENE-478064"), 
+                          (1, "Xenbase:XB-GENE-6461998"),
+                          (2, "Xenbase:XB-GENE-17342561")])
+def test_orthology_records(orthology_record, index: int, expected_subject: str):
     assert orthology_record
     association = [
         association for association in orthology_record if isinstance(association, GeneToGeneHomologyAssociation)
-    ][0]
-    assert association.subject == "Xenbase:XB-GENEPAGE-478063"
+    ][index]
+    assert association.subject == expected_subject
     assert association.predicate == "biolink:orthologous_to"
     assert association.object == "NCBIGene:8928"