Curate additional DOID-MESH mappings and contribute to DOID

scripts/add_mesh_xrefs_to_doid_owl.py

@@ -0,0 +1,141 @@
+"""This script adds newly inferred cross-references for DOID.
+
+These are directly added to the version controlled DOID OWL file.
+"""
+import csv
+
+import obonet
+
+from biomappings import load_mappings
+
+EDITABLE_OWL_PATH = "/Users/ben/src/HumanDiseaseOntology/src/ontology/doid-edit.owl"
+OBO_PATH = "/Users/ben/src/HumanDiseaseOntology/src/ontology/HumanDO.obo"
+REVIEW_PATH = "/Users/ben/src/HumanDiseaseOntology/doid_mesh_review.tsv"
+
+# Get the DOID ontology
+g = obonet.read_obo(
+    "https://raw.githubusercontent.com/DiseaseOntology/"
+    "HumanDiseaseOntology/main/src/ontology/HumanDO.obo"
+)
+
+
+def add_xref(lines, node, xref):
+    """Add xrefs to an appropriate place in the OWL file."""
+    node_owl = node.replace(":", "_")
+    look_for_xref = False
+    start_xref_idx = None
+    def_idx = None
+    xref_entries = []
+    for idx, line in enumerate(lines):
+        # First, find the class with the given ID and start looking for xrefs
+        if line.startswith("# Class: obo:%s" % node_owl):
+            blank_counter = 0
+            look_for_xref = True
+            def_idx = idx + 2
+            continue
+        # If we are already looking for xrefs after the header
+        if look_for_xref:
+            # Note that there is always a blank line right after this header, and
+            # also after the block corresponding to the entry ends. So we need
+            # to be able to tell whether we are at the first or second blank line
+            # after the header.
+            if not line.strip():
+                if not blank_counter:
+                    blank_counter += 1
+                    continue
+                else:
+                    break
+            # If we find some xrefs, we keep track of those
+            if line.startswith("AnnotationAssertion(oboInOwl:hasDbXref obo"):
+                if not start_xref_idx:
+                    start_xref_idx = idx
+                xref_entries.append(line)
+            # If we found any xrefs but now there is a different line, we finish
+            elif start_xref_idx and not line.startswith("AnnotationAssertion(oboInOwl:hasDbXref"):
+                break
+    # If we never found any existing xrefs then we will put the new xref
+    # after the definition
+    if start_xref_idx is None:
+        start_xref_idx = def_idx + 1
+    # We now have to render the xref string and sort xrefs alphabetically
+    # to make sure we put the new one in the right place
+    xref_str = 'AnnotationAssertion(oboInOwl:hasDbXref obo:%s "%s"^^xsd:string)\n' % (
+        node_owl,
+        xref,
+    )
+    xref_entries.append(xref_str)
+    xref_entries = sorted(xref_entries)
+    xr_idx = xref_entries.index(xref_str)
+    lines.insert(start_xref_idx + xr_idx, xref_str)
+    return lines
+
+
+if __name__ == "__main__":
+    # There are some curations that are redundant since DOID already mapped
+    # these nodes to MESH. We figure out what these are so we can avoid
+    # adding them.
+    doid_already_mapped = set()
+    g = obonet.read_obo(OBO_PATH)
+    for node, data in g.nodes(data=True):
+        # Skip external entries
+        if not node.startswith("DOID"):
+            continue
+        # Make sure we have a name
+        if "name" not in data:
+            continue
+        # Get existing xrefs as a standardized dict
+        xrefs = dict([xref.split(":", maxsplit=1) for xref in data.get("xref", [])])
+        # If there are already MESH mappings, we keep track of that
+        if "MESH" in xrefs:
+            doid_already_mapped.add(node)
+
+    # We now load mappings curated in Biomappings
+    mappings = load_mappings()
+    doid_mappings = [
+        (m["source identifier"], m["target identifier"], m)
+        for m in mappings
+        if (
+            m["source prefix"] == "doid"
+            and m["target prefix"] == "mesh"
+            and m["source identifier"] not in doid_already_mapped
+        )
+    ]
+    # Make sure we get and standardize the order of mappings in both directions
+    doid_mappings += [
+        (m["target identifier"], m["source identifier"], m)
+        for m in mappings
+        if (
+            m["source prefix"] == "mesh"
+            and m["target prefix"] == "doid"
+            and m["target identifier"] not in doid_already_mapped
+        )
+    ]
+
+    # Read the OWL file
+    with open(EDITABLE_OWL_PATH, "r") as fh:
+        lines = fh.readlines()
+
+    review_cols = [
+        "source prefix",
+        "source identifier",
+        "source name",
+        "relation",
+        "target prefix",
+        "target identifier",
+        "target name",
+        "type",
+        "source",
+    ]
+    review_rows = [review_cols]
+    # Add all the xrefs to the OWL, simultaneously add xrefs to a review TSV
+    for do_id, mesh_id, mapping in doid_mappings:
+        lines = add_xref(lines, do_id, "MESH:" + mesh_id)
+        review_rows.append([mapping[c] for c in review_cols])
+
+    # Dump the new review TSV and OWL file
+    with open(REVIEW_PATH, "w") as fh:
+        writer = csv.writer(fh, delimiter="\t")
+        writer.writerows(review_rows)
+
+    with open(EDITABLE_OWL_PATH, "w") as fh:
+        fh.writelines(lines)

scripts/generate_doid_mesh_mappings.py

@@ -0,0 +1,94 @@
+"""Generate mappings using Gilda from DOID to MeSH."""
+from collections import Counter
+
+import gilda
+import obonet
+from indra.databases import mesh_client
+from indra.ontology.standardize import standardize_db_refs
+from indra.tools.fix_invalidities import fix_invalidities_db_refs
+
+from biomappings import load_false_mappings, load_mappings, load_unsure
+from biomappings.resources import PredictionTuple, append_prediction_tuples
+
+# Get the DOID ontology
+g = obonet.read_obo(
+    "https://raw.githubusercontent.com/DiseaseOntology/"
+    "HumanDiseaseOntology/main/src/ontology/HumanDO.obo"
+)
+
+# Make sure we know which mappings have already been curated
+curated_mappings = set()
+for m in list(load_mappings()) + list(load_unsure()) + list(load_false_mappings()):
+    if m["source prefix"] == "doid" and m["target prefix"] == "mesh":
+        curated_mappings.add(m["source identifier"])
+    elif m["target prefix"] == "doid" and m["source prefix"] == "mesh":
+        curated_mappings.add(m["target identifier"])
+
+
+# We now iterate over all DOID entries and check for possible mappings
+mappings = {}
+existing_refs_to_mesh = set()
+already_mappable = set()
+for node, data in g.nodes(data=True):
+    # Skip external entries
+    if not node.startswith("DOID"):
+        continue
+    # Make sure we have a name
+    if "name" not in data:
+        continue
+    # Skip if already curated
+    if node in curated_mappings:
+        continue
+    # Get existing xrefs as a standardized dict
+    xrefs = [xref.split(":", maxsplit=1) for xref in data.get("xref", [])]
+    xrefs_dict = fix_invalidities_db_refs(dict(xrefs))
+    standard_refs = standardize_db_refs(xrefs_dict)
+    # If there are already MESH mappings, we keep track of that
+    if "MESH" in standard_refs:
+        already_mappable.add(node)
+    existing_refs_to_mesh |= {id for ns, id in standard_refs.items() if ns == "MESH"}
+    # We can now ground the name and specifically look for MESH matches
+    matches = gilda.ground(data["name"], namespaces=["MESH"])
+    # If we got a match, we add the MESH ID as a mapping
+    if matches:
+        for grounding in matches[0].get_groundings():
+            if grounding[0] == "MESH":
+                mappings[node] = matches[0].term.id
+
+
+print("Found %d DOID->MESH mappings." % len(mappings))
+
+# We makes sure that (i) the node is not already mappable to MESH and that
+# (ii) there isn't some other node that was not already mapped to the
+# given MESH ID
+mappings = {
+    k: v
+    for k, v in mappings.items()
+    if v not in existing_refs_to_mesh and k not in already_mappable
+}
+
+# We now need to make sure that we don't reuse the same MESH ID across
+# multiple predicted mappings
+cnt = Counter(mappings.values())
+mappings = {k: v for k, v in mappings.items() if cnt[v] == 1}
+
+print("Found %d filtered DOID->MESH mappings." % len(mappings))
+
+# We can now add the predictions
+predictions = []
+for doid, mesh_id in mappings.items():
+    pred = PredictionTuple(
+        target_prefix="doid",
+        target_identifier=doid[5:],
+        target_name=g.nodes[doid]["name"],
+        relation="skos:exactMatch",
+        source_prefix="mesh",
+        source_id=mesh_id,
+        source_name=mesh_client.get_mesh_name(mesh_id),
+        type="lexical",
+        confidence=0.9,
+        source="generate_doid_mesh_mappings.py",
+    )
+    predictions.append(pred)
+
+append_prediction_tuples(predictions, deduplicate=True, sort=True)

setup.cfg

@@ -50,7 +50,7 @@ install_requires =
     pyyaml
     tqdm
     pystow>=0.2.7
-    bioregistry>=0.4.29
+    bioregistry>=0.5.104
 
 zip_safe = false
 include_package_data = True

src/biomappings/resources/incorrect.tsv

@@ -132,9 +132,12 @@ doid	DOID:0050575	D-2-hydroxyglutaric aciduria	skos:exactMatch	mesh	C535306	2-Hy
 doid	DOID:0050575	D-2-hydroxyglutaric aciduria	skos:exactMatch	umls	C2746066	Combined D-2- and L-2-hydroxyglutaric aciduria	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0050577	cranioectodermal dysplasia	skos:exactMatch	umls	C0432235	CRANIOECTODERMAL DYSPLASIA 1	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0050580	hereditary lymphedema	skos:exactMatch	mesh	D008209	Lymphedema	manually_reviewed	orcid:0000-0001-9439-5346
+doid	DOID:0050650	familial atrial fibrillation	skos:exactMatch	mesh	D001281	Atrial Fibrillation	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0050665	fetal alcohol syndrome	skos:exactMatch	mesh	D063647	Fetal Alcohol Spectrum Disorders	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0050666	partial fetal alcohol syndrome	skos:exactMatch	mesh	D063647	Fetal Alcohol Spectrum Disorders	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0050667	alcohol-related neurodevelopmental disorder	skos:exactMatch	mesh	D063647	Fetal Alcohol Spectrum Disorders	manually_reviewed	orcid:0000-0001-9439-5346
+doid	DOID:0050672	dyskinetic cerebral palsy	skos:exactMatch	mesh	D002547	Cerebral Palsy	manually_reviewed	orcid:0000-0001-9439-5346
+doid	DOID:0050673	mixed cerebral palsy	skos:exactMatch	mesh	D002547	Cerebral Palsy	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0050685	small cell carcinoma	skos:exactMatch	umls	C0149925	Small cell carcinoma of lung	manually_reviewed	orcid:0000-0003-4423-4370
 doid	DOID:0050882	spinocerebellar ataxia type 5	skos:exactMatch	mesh	D020754	Spinocerebellar Ataxias	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0050954	spinocerebellar ataxia type 1	skos:exactMatch	mesh	D020754	Spinocerebellar Ataxias	manually_reviewed	orcid:0000-0001-9439-5346
@@ -148,11 +151,15 @@ doid	DOID:0060141	finger agnosia	skos:exactMatch	mesh	D000377	Agnosia	manually_r
 doid	DOID:0060151	tactile agnosia	skos:exactMatch	mesh	D000377	Agnosia	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0060152	time agnosia	skos:exactMatch	mesh	D000377	Agnosia	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0060153	topographical agnosia	skos:exactMatch	mesh	D000377	Agnosia	manually_reviewed	orcid:0000-0001-9439-5346
+doid	DOID:0070197	distal myopathy 1	skos:exactMatch	mesh	D049310	Distal Myopathies	manually_reviewed	orcid:0000-0001-9439-5346
+doid	DOID:0070340	classic citrullinemia	skos:exactMatch	mesh	D020159	Citrullinemia	manually_reviewed	orcid:0000-0001-9439-5346
+doid	DOID:0080046	Stickler syndrome	skos:exactMatch	mesh	C537492	Stickler syndrome, type 1	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0080333	aortic valve disease 1	skos:exactMatch	mesh	D000082882	Bicuspid Aortic Valve Disease	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0080546	non-alcoholic fatty liver	skos:exactMatch	mesh	D065626	Non-alcoholic Fatty Liver Disease	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0080547	non-alcoholic steatohepatitis	skos:exactMatch	mesh	D065626	Non-alcoholic Fatty Liver Disease	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0080748	chronic inducible urticaria	skos:exactMatch	mesh	D000080223	Chronic Urticaria	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0080749	chronic spontaneous urticaria	skos:exactMatch	mesh	D000080223	Chronic Urticaria	manually_reviewed	orcid:0000-0001-9439-5346
+doid	DOID:0080939	hereditary angioedema type I	skos:exactMatch	mesh	D056829	Hereditary Angioedema Types I and II	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0111135	congenital generalized lipodystrophy type 1	skos:exactMatch	mesh	D052497	Lipodystrophy, Congenital Generalized	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:0111136	congenital generalized lipodystrophy type 2	skos:exactMatch	mesh	D052497	Lipodystrophy, Congenital Generalized	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:1024	leprosy	skos:exactMatch	umls	C0051981	leprosy vaccine	manually_reviewed	orcid:0000-0003-4423-4370
@@ -170,6 +177,7 @@ doid	DOID:1498	cholera	skos:exactMatch	umls	C0008359	cholera vaccine	manually_re
 doid	DOID:1586	rheumatic fever	skos:exactMatch	umls	C1548484	rheumatic fever vaccine	manually_reviewed	orcid:0000-0003-4423-4370
 doid	DOID:1909	melanoma	skos:exactMatch	umls	C0796561	Melanoma vaccine	manually_reviewed	orcid:0000-0003-4423-4370
 doid	DOID:2043	hepatitis B	skos:exactMatch	umls	C0062527	hepatitis B surface antigen vaccine	manually_reviewed	orcid:0000-0003-4423-4370
+doid	DOID:3479	uveal cancer	skos:exactMatch	mesh	D014604	Uveal Neoplasms	manually_reviewed	orcid:0000-0001-9439-5346
 doid	DOID:3482	plague	skos:exactMatch	umls	C0032066	Plague Vaccine	manually_reviewed	orcid:0000-0003-4423-4370
 doid	DOID:3910	lung adenocarcinoma	skos:exactMatch	efo	0005288	non-small cell lung adenocarcinoma	manually_reviewed	orcid:0000-0003-4423-4370
 doid	DOID:4552	large cell carcinoma	skos:exactMatch	efo	0003050	large cell lung carcinoma	manually_reviewed	orcid:0000-0003-4423-4370
@@ -358,6 +366,7 @@ mesh	C535442	Bile acid synthesis defect, congenital, 1	skos:exactMatch	doid	DOID
 mesh	C535536	Iridogoniodysgenesis, dominant type	skos:exactMatch	doid	DOID:0050786	iridogoniodysgenesis syndrome	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	C535837	Pancreatic carcinoma, familial	skos:exactMatch	doid	DOID:4905	pancreatic carcinoma	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	C535847	Hay-Wells syndrome	skos:exactMatch	doid	DOID:0090119	ankyloblepharon-ectodermal defects-cleft lip/palate syndrome	manually_reviewed	orcid:0000-0003-1307-2508
+mesh	C536133	Meckel syndrome type 1	skos:exactMatch	doid	DOID:0050778	Meckel syndrome	manually_reviewed	orcid:0000-0001-9439-5346
 mesh	C536192	Brittle cornea syndrome 1	skos:exactMatch	doid	DOID:14775	brittle cornea syndrome	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	C536277	Idiopathic dilation cardiomyopathy	skos:exactMatch	doid	DOID:0110429	dilated cardiomyopathy 1H	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	C536300	Partington X-linked mental retardation syndrome	skos:exactMatch	doid	DOID:14744	Partington syndrome	manually_reviewed	orcid:0000-0003-1307-2508
@@ -379,7 +388,6 @@ mesh	C567200	Immunodeficiency, Hypogammaglobulinemia, and Reduced B Cells	skos:e
 mesh	C567484	Mental Retardation, X-Linked, Syndromic, Christianson Type	skos:exactMatch	doid	DOID:0060825	Christianson syndrome	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	C567738	Neuropathy, Hereditary Sensory And Autonomic, Type IIB	skos:exactMatch	doid	DOID:0070150	hereditary sensory and autonomic neuropathy type 2B	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	C579934	Autosomal Recessive Cerebellar Ataxia Type 1	skos:exactMatch	doid	DOID:0111618	autosomal recessive spinocerebellar ataxia 8	manually_reviewed	orcid:0000-0003-1307-2508
-mesh	C580012	Congenital Fibrosis of the Extraocular Muscles	skos:exactMatch	doid	DOID:0080143	congenital fibrosis of the extraocular muscles	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	C580473	Succinate-Coa Ligase Deficiency	skos:exactMatch	doid	DOID:0080124	mitochondrial DNA depletion syndrome 5	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	C584135	endotrophin	skos:exactMatch	ncit	C165979	Pro-C6 Measurement	manually_reviewed	orcid:0000-0001-9439-5346
 mesh	D000071636	Protein Phosphatase 2C	skos:exactMatch	hgnc	9279	PDP1	manually_reviewed	orcid:0000-0001-9439-5346
@@ -413,6 +421,7 @@ mesh	D003924	Diabetes Mellitus, Type 2	skos:exactMatch	doid	DOID:0050524	maturit
 mesh	D003925	Diabetic Angiopathies	skos:exactMatch	doid	DOID:10182	diabetic peripheral angiopathy	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	D004407	Dysgerminoma	skos:exactMatch	doid	DOID:5511	dysgerminoma of ovary	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	D004410	Dyslexia	skos:exactMatch	doid	DOID:13365	reading disorder	manually_reviewed	orcid:0000-0003-1307-2508
+mesh	D004697	Endocarditis, Bacterial	skos:exactMatch	doid	DOID:0060000	infective endocarditis	manually_reviewed	orcid:0000-0001-9439-5346
 mesh	D004932	Esophageal and Gastric Varices	skos:exactMatch	doid	DOID:112	esophageal varix	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	D005319	Fetal Hemoglobin	skos:exactMatch	efo	0004576	fetal hemoglobin measurement	manually_reviewed	orcid:0000-0003-4423-4370
 mesh	D005351	Fibromatosis, Gingival	skos:exactMatch	doid	DOID:0060466	gingival fibromatosis	manually_reviewed	orcid:0000-0003-1307-2508
@@ -424,7 +433,6 @@ mesh	D006441	Hemoglobin A	skos:exactMatch	efo	0009208	Hemoglobin A Measurement	m
 mesh	D006443	Hemoglobin A2	skos:exactMatch	efo	0005845	hemoglobin A2 measurement	manually_reviewed	orcid:0000-0003-4423-4370
 mesh	D006451	Hemoglobin, Sickle	skos:exactMatch	efo	0009223	Hemoglobin S Measurement	manually_reviewed	orcid:0000-0003-4423-4370
 mesh	D006480	Hemorrhagic Fever with Renal Syndrome	skos:exactMatch	doid	DOID:0050200	Korean hemorrhagic fever	manually_reviewed	orcid:0000-0003-1307-2508
-mesh	D006519	Hepatitis, Alcoholic	skos:exactMatch	doid	DOID:12351	alcoholic hepatitis	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	D006593	Hexokinase	skos:exactMatch	go	GO:0004396	hexokinase activity	manually_reviewed	orcid:0000-0001-9439-5346
 mesh	D006646	Histiocytosis, Langerhans-Cell	skos:exactMatch	doid	DOID:2571	Langerhans-cell histiocytosis	manually_reviewed	orcid:0000-0003-1307-2508
 mesh	D007014	Hypophosphatasia	skos:exactMatch	doid	DOID:0110915	childhood hypophosphatasia	manually_reviewed	orcid:0000-0003-1307-2508