Merge pull request #86 from biopragmatics/uberon_mesh

Add UBERON-MeSH mappings

Author: bgyori

scripts/generate_uberon_mesh_mappings.py

@@ -0,0 +1,40 @@
+"""Generate mappings using Gilda from UBERON to MeSH."""
+
+import gilda
+import obonet
+from indra.databases import mesh_client
+
+from biomappings.resources import PredictionTuple, append_prediction_tuples
+
+g = obonet.read_obo("/Users/ben/src/uberon/src/ontology/uberon-edit.obo")
+
+mappings = {}
+for node, data in g.nodes(data=True):
+    if not node.startswith("UBERON"):
+        continue
+    mesh_refs = [xref for xref in data.get("xref", []) if xref.startswith("MESH")]
+    if mesh_refs:
+        continue
+    matches = gilda.ground(data["name"])
+    if matches and matches[0].term.db == "MESH":
+        mappings[node] = matches[0].term.id
+
+print("Found %d UBERON->MESH mappings." % len(mappings))
+
+predictions = []
+for uberon_id, mesh_id in mappings.items():
+    pred = PredictionTuple(
+        source_prefix="uberon",
+        source_id=uberon_id,
+        source_name=g.nodes[uberon_id]["name"],
+        relation="skos:exactMatch",
+        target_prefix="mesh",
+        target_identifier=mesh_id,
+        target_name=mesh_client.get_mesh_name(mesh_id),
+        type="lexical",
+        confidence=0.9,
+        source="generate_uberon_mesh_mappings.py",
+    )
+    predictions.append(pred)
+
+append_prediction_tuples(predictions, deduplicate=True, sort=True)

src/biomappings/resources/incorrect.tsv

@@ -938,6 +938,9 @@ reactome	R-XTR-8877627	Vitamin E	skos:exactMatch	chebi	CHEBI:46430	(-)-alpha-toc
 reactome	R-XTR-8877627	Vitamin E	skos:exactMatch	mesh	D014810	Vitamin E	manually_reviewed	orcid:0000-0003-4423-4370
 reactome	R-XTR-9664873	Pexophagy	speciesSpecific	go	GO:0030242	autophagy of peroxisome	manually_reviewed	orcid:0000-0003-4423-4370
 reactome	R-XTR-983189	Kinesins	speciesSpecific	fplx	Kinesin	Kinesin	manually_reviewed	orcid:0000-0003-4423-4370
+uberon	UBERON:0000104	life cycle	skos:exactMatch	mesh	D008018	Life Cycle Stages	manually_reviewed	orcid:0000-0001-9439-5346
+uberon	UBERON:0022469	primary olfactory cortex	skos:exactMatch	mesh	D066194	Olfactory Cortex	manually_reviewed	orcid:0000-0001-9439-5346
+uberon	UBERON:2001977	pad	skos:exactMatch	mesh	D058729	Peripheral Arterial Disease	manually_reviewed	orcid:0000-0001-9439-5346
 umls	C0006142	Malignant neoplasm of breast	skos:exactMatch	mesh	D001943	Breast Neoplasms	manual	orcid:0000-0002-6601-2165
 wikipathways	WP1060	Matrix Metalloproteinases	speciesSpecific	fplx	MMP	MMP	manually_reviewed	orcid:0000-0003-4423-4370
 wikipathways	WP1176	Matrix Metalloproteinases	speciesSpecific	fplx	MMP	MMP	manually_reviewed	orcid:0000-0003-4423-4370