Generate and curate a new set of MESH-CHEBI mappings

scripts/generate_chebi_mesh_mappings.py

@@ -0,0 +1,60 @@
+"""Generate mappings using Gilda from UBERON to MeSH."""
+
+from collections import Counter
+
+from gilda.api import grounder
+from indra.ontology.bio import bio_ontology
+
+from biomappings.resources import PredictionTuple, append_prediction_tuples
+
+
+if __name__ == "__main__":
+    grounder.ground("x")
+    ambigs = {
+        k: v for k, v in grounder.grounder.entries.items() if len({(vv.db, vv.id) for vv in v}) > 1
+    }
+    mesh_chebi = [
+        v for v in ambigs.values() if len(v) == 2 and {vv.db for vv in v} == {"MESH", "CHEBI"}
+    ]
+    entries = []
+    for mesh_chebi_pair in mesh_chebi:
+        for term in mesh_chebi_pair:
+            entries.append((term.db, term.id))
+    cnt = Counter(entries)
+    mesh_chebi_simple = []
+    for mesh_chebi_pair in mesh_chebi:
+        if any(cnt[(term.db, term.id)] > 1 for term in mesh_chebi_pair):
+            continue
+        mesh_chebi_simple.append(mesh_chebi_pair)
+
+    print("Found %d CHEBI-MESH mappings." % len(mesh_chebi_simple))
+
+    predictions = []
+    n_redundant = 0
+    for pair in mesh_chebi_simple:
+        chebi_term = [term for term in pair if term.db == "CHEBI"][0]
+        mesh_term = [term for term in pair if term.db == "MESH"][0]
+
+        mappings = bio_ontology.get_mappings("MESH", mesh_term.id)
+        if ("CHEBI", chebi_term.id) in mappings:
+            n_redundant += 1
+
+        pred = PredictionTuple(
+            source_prefix="chebi",
+            source_id=chebi_term.id,
+            source_name=chebi_term.entry_name,
+            relation="skos:exactMatch",
+            target_prefix="mesh",
+            target_identifier=mesh_term.id,
+            target_name=mesh_term.entry_name,
+            type="lexical",
+            confidence=0.95,
+            source="generate_chebi_mesh_mappings.py",
+        )
+        predictions.append(pred)
+
+    print(
+        "A total of %d mappings could be indirectly inferred from"
+        "INDRA ontology xrefs" % len(n_redundant)
+    )
+    append_prediction_tuples(predictions, deduplicate=True, sort=True)

src/biomappings/resources/incorrect.tsv

@@ -35,6 +35,8 @@ ccle	PC3_PROSTATE	PC-3	skos:exactMatch	cellosaurus	CVCL_S982	PC-3 [Human lung ca
 ccle	PL21_HAEMATOPOIETIC_AND_LYMPHOID_TISSUE	PL-21	skos:exactMatch	cellosaurus	CVCL_E295	TS 0129	manually_reviewed	orcid:0000-0001-9439-5346
 ccle	TC71_BONE	TC-71	skos:exactMatch	cellosaurus	CVCL_S882	TC71	manually_reviewed	orcid:0000-0003-4423-4370
 ccle	WM239A	WM239A	skos:exactMatch	cellosaurus	CVCL_RN56	WM239B	manually_reviewed	orcid:0000-0003-4423-4370
+chebi	CHEBI:135025	rimazolium	skos:exactMatch	mesh	C100296	probon	manually_reviewed	orcid:0000-0001-9439-5346
+chebi	CHEBI:135083	cibenzoline	skos:exactMatch	mesh	C032151	cifenline	manually_reviewed	orcid:0000-0001-9439-5346
 chebi	CHEBI:17544	hydrogencarbonate	skos:exactMatch	mesh	D001639	Bicarbonates	manually_reviewed	orcid:0000-0003-4423-4370
 chebi	CHEBI:29806	fumarate(2-)	skos:exactMatch	mesh	D005650	Fumarates	manually_reviewed	orcid:0000-0003-4423-4370
 chebi	CHEBI:33224	chromane	skos:exactMatch	mesh	D002839	Chromans	manually_reviewed	orcid:0000-0003-4423-4370
@@ -363,6 +365,7 @@ mesh	D007161	Immunoproliferative Small Intestinal Disease	skos:exactMatch	doid	D
 mesh	D007319	Sleep Initiation and Maintenance Disorders	skos:exactMatch	doid	DOID:0111141	delayed sleep phase syndrome	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	D007588	Job Satisfaction	skos:exactMatch	efo	0009723	job satisfaction measurement	manually_reviewed	orcid:0000-0003-4423-4370
 mesh	D007589	Job Syndrome	skos:exactMatch	doid	DOID:0080545	hyper IgE syndrome	manually_reviewed	orcid:0000-0003-1307-2508
+mesh	D007605	Juvenile Hormones	skos:exactMatch	chebi	CHEBI:24851	insect growth regulator	manually_reviewed	orcid:0000-0001-9439-5346
 mesh	D007680	Kidney Neoplasms	skos:exactMatch	doid	DOID:4451	renal carcinoma	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	D007713	Klinefelter Syndrome	skos:exactMatch	doid	DOID:0090070	hypogonadotropic hypogonadism	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	D007762	Labyrinthitis	skos:exactMatch	doid	DOID:3930	otitis interna	manually_reviewed	orcid:0000-0003-1307-2508
@@ -476,6 +479,7 @@ mesh	D051057	Proto-Oncogene Proteins c-akt	skos:exactMatch	hgnc	391	AKT1	manual
 mesh	D051236	Receptors, Pituitary Adenylate Cyclase-Activating Polypeptide	skos:exactMatch	go	GO:0001634	pituitary adenylate cyclase-activating polypeptide receptor activity	manually_reviewed	orcid:0000-0003-4423-4370
 mesh	D052003	NF-kappa B p52 Subunit	skos:exactMatch	hgnc	7795	NFKB2	manually_reviewed	orcid:0000-0003-4423-4370
 mesh	D053483	Eye Movement Measurements	skos:exactMatch	efo	0007699	eye movement measurement	manually_reviewed	orcid:0000-0003-4423-4370
+mesh	D053496	Inositol 1,4,5-Trisphosphate Receptors	skos:exactMatch	chebi	CHEBI:131186	IP3 receptor antagonist	manually_reviewed	orcid:0000-0001-9439-5346
 mesh	D053840	Brugada Syndrome	skos:exactMatch	doid	DOID:0110218	Brugada syndrome 1	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	D054000	Nevus, Sebaceous of Jadassohn	skos:exactMatch	doid	DOID:7039	Borst-Jadassohn intraepidermal carcinoma	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	D054467	Phospholipases A2	skos:exactMatch	hgnc	9030	PLA2G1B	manually_reviewed	orcid:0000-0001-9439-5346