add impc x gwas config

EBISPOT · Nov 6, 2024 · 41d0a9c · 41d0a9c
1 parent 0e08f86
commit 41d0a9c
Show file tree

Hide file tree

Showing 4 changed files with 182 additions and 1 deletion.
diff --git a/configs/datasource_configs/ols_impc_x_gwas.yaml b/configs/datasource_configs/ols_impc_x_gwas.yaml
@@ -0,0 +1,9 @@
+name: OLS
+enabled: true
+ingests:
+  - globs: ["/nfs/production/parkinso/spot/ols4/prod/slurm_pipeline/ontologies.json.gz"]
+    command: '
+      grebi_ingest_ols
+        --datasource-name $GREBI_INGEST_DATASOURCE_NAME
+        --ontologies efo,mp,hp,ro,iao,uberon,pato,oba,bspo,iao,obi,bfo,cob,oio,iao,biolink,mondo,doid,dc,dcterms,ncit,edam,upheno
+        --skip-obsolete'
diff --git a/configs/subgraph_configs/Makefile b/configs/subgraph_configs/Makefile
@@ -1,6 +1,6 @@
 
 
-all: ebi_monarch_xspecies.json ebi_monarch.json hett.json hra_kg.json impc.json monarch.json
+all: ebi_monarch_xspecies.json ebi_monarch.json hett.json hra_kg.json impc.json monarch.json impc_x_gwas.json
 
 %.json: src/%.py
 	python3 $< > $@

diff --git a/configs/subgraph_configs/impc_x_gwas.json b/configs/subgraph_configs/impc_x_gwas.json
@@ -0,0 +1,145 @@
+{
+  "id": "IMPC x GWAS",
+  "name": "IMPC and the GWAS Catalog",
+  "bytes_per_merged_file": 1073741824,
+  "identifier_props": [
+    "id",
+    "owl:equivalentClass",
+    "owl:equivalentProperty",
+    "owl:sameAs",
+    "grebi:hashId",
+    "grebi:equivalentTo",
+    "ols:iri",
+    "ols:shortForm",
+    "hgnc:ensembl_gene_id",
+    "obo:chebi/inchi",
+    "obo:chebi/inchikey",
+    "obo:chebi/smiles",
+    "impc:pmId",
+    "impc:humanGeneAccId",
+    "monarch:iri",
+    "skos:exactMatch",
+    "ncit:P368",
+    "ncit:C98965",
+    "dcterms:identifier",
+    "oboinowl:hasAlternativeId",
+    "robokop:equivalent_identifiers",
+    "mesh.vocab:identifier"
+  ],
+  "type_superclasses": [
+    "mondo:0000001",
+    "efo:0000408",
+    "chebi:36080",
+    "chebi:24431"
+  ],
+  "additional_equivalence_groups": [
+    [
+      "grebi:name",
+      "ols:label",
+      "rdfs:label",
+      "monarch:name",
+      "impc:name",
+      "reactome:displayName",
+      "dcterms:title",
+      "ncit:Preferred_Name",
+      "robokop:name"
+    ],
+    [
+      "grebi:description",
+      "iao:definition",
+      "monarch:description",
+      "ols:definition",
+      "robokop:description"
+    ],
+    [
+      "grebi:synonym",
+      "monarch:synonym",
+      "iao:alternative_label",
+      "ols:synonym",
+      "oboinowl:hasExactSynonym",
+      "dcterms:alternative"
+    ],
+    [
+      "mondo:0000001",
+      "ogms:0000031"
+    ],
+    [
+      "biolink:broad_match",
+      "skos:broader",
+      "skos:broadMatch",
+      "ols:directAncestor"
+    ],
+    [
+      "biolink:subclass_of",
+      "ols:directParent",
+      "rdfs:subClassOf",
+      "rdfs:subPropertyOf"
+    ],
+    [
+      "rdfs:isDefinedBy",
+      "ols:ontologyIri",
+      "ols:ontologyId"
+    ]
+  ],
+  "exclude_props": [
+    "ols:hierarchicalProperty",
+    "ols:synonymProperty",
+    "ols:curie",
+    "ols:shortForm",
+    "ols:ontologyPreferredPrefix",
+    "ols:iri",
+    "ols:uri",
+    "ols:imported",
+    "ols:hasHierarchicalParents",
+    "ols:hasHierarchicalChildren",
+    "ols:hasDirectParents",
+    "ols:hasDirectChildren",
+    "ols:numDescendants",
+    "ols:numHierarchicalDescendants",
+    "oboinowl:id",
+    "oboinowl:url",
+    "monarch:iri",
+    "cco:hasDocument",
+    "cco:hasMolecule"
+  ],
+  "exclude_edges": [],
+  "exclude_self_referential_edges": [
+    "foaf:page",
+    "rdfs:seeAlso",
+    "oboinowl:hasDbXref",
+    "biolink:subclass_of",
+    "biolink:broad_match",
+    "ols:ontology_purl",
+    "ols:ontologyId",
+    "cheminf:000407",
+    "cheminf:InChIKey",
+    "biolink:interacts_with",
+    "reactome:url",
+    "dc:Identifier",
+    "hgnc:agr",
+    "hgnc:gencc",
+    "monarch:xref",
+    "hgnc:uniprot_ids",
+    "hgnc:omim_id",
+    "hgnc:entrez_id",
+    "reactome:referenceGene",
+    "reactome:identifier",
+    "reactome:crossReference",
+    "edam:Ensembl_gene_id",
+    "obo:pr#has_gene_template",
+    "ols:relatedTo",
+    "ols:relatedFrom",
+    "robokop:smiles",
+    "ctd:ChemicalURL",
+    "ctd:UniProtIDs"
+  ],
+  "datasource_configs": [
+    "./configs/datasource_configs/impc.yaml",
+    "./configs/datasource_configs/gwas.yaml",
+    "./configs/datasource_configs/ols.yaml",
+    "./configs/datasource_configs/mondo_efo.yaml",
+    "./configs/datasource_configs/monarch.yaml",
+    "./configs/datasource_configs/sssom.yaml",
+    "./configs/datasource_configs/mondo_efo.yaml"
+  ]
+}
diff --git a/configs/subgraph_configs/src/impc_x_gwas.py b/configs/subgraph_configs/src/impc_x_gwas.py
@@ -0,0 +1,27 @@
+
+
+from shared import config
+
+config['id'] = 'IMPC x GWAS'
+config['name'] = 'IMPC and the GWAS Catalog'
+config['datasource_configs'] = [
+    "./configs/datasource_configs/impc.yaml",
+    "./configs/datasource_configs/gwas.yaml",
+    "./configs/datasource_configs/ols.yaml",
+    "./configs/datasource_configs/mondo_efo.yaml",
+    "./configs/datasource_configs/monarch.yaml",
+    "./configs/datasource_configs/sssom.yaml",
+    "./configs/datasource_configs/mondo_efo.yaml"
+]
+config['identifier_props'] = [
+    "id",
+    "monarch:iri",
+    "ols:iri",
+    "ols:shortForm",
+    "impc:pmId",
+    "impc:humanGeneAccId"
+]
+
+if __name__ == '__main__':
+    import json
+    print(json.dumps(config, indent=2))