First draft ncbigene and fix hgnc import (#7380)

* First draft ncbigene import * Changes the HGNC ingest to use the symbol as the label --------- Co-authored-by: Sabrina Toro <[email protected]>
monarch-initiative · Mar 30, 2024 · 6ec68b6 · 6ec68b6
1 parent d7df62e
commit 6ec68b6
Show file tree

Hide file tree

Showing 5 changed files with 67 additions and 26 deletions.
diff --git a/src/ontology/Makefile b/src/ontology/Makefile
@@ -71,7 +71,7 @@ all: test all_artefacts mappings
 # Artefacts
 # ----------------------------------------
 
-IMPORTS =  uberon cl go pato envo ro hp mf ncbitaxon chebi ncit hgnc foodon so ecto omo chr hsapdv nbo maxo mfomd
+IMPORTS =  uberon cl go pato envo ro hp mf ncbitaxon chebi ncit hgnc foodon so ecto omo chr hsapdv nbo maxo mfomd ncbigene
 
 IMPORT_ROOTS =  $(IMPORTDIR)/merged_import
 IMPORT_OWL_FILES = $(foreach n,$(IMPORT_ROOTS), $(n).owl)
@@ -678,15 +678,25 @@ $(MIRRORDIR)/%.owl: mirror-% | $(MIRRORDIR)
 
 ###### END PARTIAL ODK MIGRATION ##############
 ###### BEGIN CUSTOM IMPORTS ###################
-mirror/dipper-%.ttl: | $(MIRRORDIR)
-	if [ $(MIR) = true ] && [ $(IMP) = true ]; then wget --no-check-certificate https://archive.monarchinitiative.org/latest/rdf/$*.ttl -O $@.tmp &&\
+mirror/hgnc_gene.nt: | $(MIRRORDIR)
+	if [ $(MIR) = true ] && [ $(IMP) = true ]; then curl -L https://data.monarchinitiative.org/monarch-kg/latest/rdf/hgnc_gene.nt.gz | gzip -d > $@.tmp &&\
 		perl -npe 's@https://www.genenames.org/data/gene-symbol-report/#!/hgnc_id/HGNC:@http://identifiers.org/hgnc/@g' $@.tmp > $@; fi
-.PRECIOUS: mirror/dipper-%.ttl
+.PRECIOUS: mirror/hgnc_gene.nt
 
-mirror-hgnc: mirror/dipper-hgnc.ttl | $(TMPDIR)
+mirror/ncbi_gene.nt: | $(MIRRORDIR)
+	if [ $(MIR) = true ] && [ $(IMP) = true ]; then curl -L https://data.monarchinitiative.org/monarch-kg/latest/rdf/ncbi_gene.nt.gz | gzip -d > $@.tmp &&\
+		perl -npe 's@https://www.genenames.org/data/gene-symbol-report/#!/hgnc_id/HGNC:@http://identifiers.org/hgnc/@g' $@.tmp > $@; fi
+.PRECIOUS: mirror/ncbi_gene.nt
+
+mirror-hgnc: mirror/hgnc_gene.nt | $(TMPDIR)
 	if [ $(MIR) = true ] && [ $(IMP) = true ]; then $(ROBOT) merge -i $< \
 		query --format ttl --query ../sparql/construct/construct-hgnc.sparql $(TMPDIR)/$@.owl; fi
 
+mirror-ncbigene: mirror/ncbi_gene.nt | $(TMPDIR)
+	if [ $(MIR) = true ] && [ $(IMP) = true ]; then $(ROBOT) merge -i $< \
+		query --format ttl --query ../sparql/construct/construct-ncbigene.sparql $(TMPDIR)/$@.owl; fi
+
+
 IMPORTSEED=$(IMPORTDIR)/seed.txt # SHOULD BE TMP
 ###### END CUSTOM IMPORTS ###################
 

diff --git a/src/ontology/imports/merged_import.owl b/src/ontology/imports/merged_import.owl
@@ -27776,9 +27776,9 @@ SubClassOf(<http://identifiers.org/hgnc/10892> <http://purl.obolibrary.org/obo/S
 AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/10896> "SKI")
 SubClassOf(<http://identifiers.org/hgnc/10896> <http://purl.obolibrary.org/obo/SO_0000704>)
 
-# Class: <http://identifiers.org/hgnc/10898> (SKIV2L)
+# Class: <http://identifiers.org/hgnc/10898> (SKIC2)
 
-AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/10898> "SKIV2L")
+AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/10898> "SKIC2")
 SubClassOf(<http://identifiers.org/hgnc/10898> <http://purl.obolibrary.org/obo/SO_0000704>)
 
 # Class: <http://identifiers.org/hgnc/1090> (DST)
@@ -28106,9 +28106,9 @@ SubClassOf(<http://identifiers.org/hgnc/11071> <http://purl.obolibrary.org/obo/S
 AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/11073> "SLC9A3")
 SubClassOf(<http://identifiers.org/hgnc/11073> <http://purl.obolibrary.org/obo/SO_0000704>)
 
-# Class: <http://identifiers.org/hgnc/11075> (SLC9A3R1)
+# Class: <http://identifiers.org/hgnc/11075> (NHERF1)
 
-AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/11075> "SLC9A3R1")
+AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/11075> "NHERF1")
 SubClassOf(<http://identifiers.org/hgnc/11075> <http://purl.obolibrary.org/obo/SO_0000704>)
 
 # Class: <http://identifiers.org/hgnc/11079> (SLC9A6)
@@ -32716,9 +32716,9 @@ SubClassOf(<http://identifiers.org/hgnc/1810> <http://purl.obolibrary.org/obo/SO
 AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/18111> "CCDC50")
 SubClassOf(<http://identifiers.org/hgnc/18111> <http://purl.obolibrary.org/obo/SO_0000704>)
 
-# Class: <http://identifiers.org/hgnc/18119> (SPATA5)
+# Class: <http://identifiers.org/hgnc/18119> (AFG2A)
 
-AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/18119> "SPATA5")
+AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/18119> "AFG2A")
 SubClassOf(<http://identifiers.org/hgnc/18119> <http://purl.obolibrary.org/obo/SO_0000704>)
 
 # Class: <http://identifiers.org/hgnc/18121> (MFRP)
@@ -33391,9 +33391,9 @@ SubClassOf(<http://identifiers.org/hgnc/19087> <http://purl.obolibrary.org/obo/S
 AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/19100> "IL23R")
 SubClassOf(<http://identifiers.org/hgnc/19100> <http://purl.obolibrary.org/obo/SO_0000704>)
 
-# Class: <http://identifiers.org/hgnc/19102> (DDX58)
+# Class: <http://identifiers.org/hgnc/19102> (RIGI)
 
-AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/19102> "DDX58")
+AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/19102> "RIGI")
 SubClassOf(<http://identifiers.org/hgnc/19102> <http://purl.obolibrary.org/obo/SO_0000704>)
 
 # Class: <http://identifiers.org/hgnc/19104> (NPHP4)
@@ -35351,9 +35351,9 @@ SubClassOf(<http://identifiers.org/hgnc/236> <http://purl.obolibrary.org/obo/SO_
 AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/23631> "NPSR1")
 SubClassOf(<http://identifiers.org/hgnc/23631> <http://purl.obolibrary.org/obo/SO_0000704>)
 
-# Class: <http://identifiers.org/hgnc/23639> (TTC37)
+# Class: <http://identifiers.org/hgnc/23639> (SKIC3)
 
-AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/23639> "TTC37")
+AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/23639> "SKIC3")
 SubClassOf(<http://identifiers.org/hgnc/23639> <http://purl.obolibrary.org/obo/SO_0000704>)
 
 # Class: <http://identifiers.org/hgnc/2364> (CRLF1)
@@ -35736,9 +35736,9 @@ SubClassOf(<http://identifiers.org/hgnc/24576> <http://purl.obolibrary.org/obo/S
 AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/24579> "CIB2")
 SubClassOf(<http://identifiers.org/hgnc/24579> <http://purl.obolibrary.org/obo/SO_0000704>)
 
-# Class: <http://identifiers.org/hgnc/24587> (FAM126A)
+# Class: <http://identifiers.org/hgnc/24587> (HYCC1)
 
-AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/24587> "FAM126A")
+AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/24587> "HYCC1")
 SubClassOf(<http://identifiers.org/hgnc/24587> <http://purl.obolibrary.org/obo/SO_0000704>)
 
 # Class: <http://identifiers.org/hgnc/24592> (STEAP3)
@@ -40181,9 +40181,9 @@ SubClassOf(<http://identifiers.org/hgnc/4174> <http://purl.obolibrary.org/obo/SO
 AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/4175> "GATM")
 SubClassOf(<http://identifiers.org/hgnc/4175> <http://purl.obolibrary.org/obo/SO_0000704>)
 
-# Class: <http://identifiers.org/hgnc/4177> (GBA)
+# Class: <http://identifiers.org/hgnc/4177> (GBA1)
 
-AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/4177> "GBA")
+AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/4177> "GBA1")
 SubClassOf(<http://identifiers.org/hgnc/4177> <http://purl.obolibrary.org/obo/SO_0000704>)
 
 # Class: <http://identifiers.org/hgnc/4180> (GBE1)
@@ -44971,9 +44971,9 @@ SubClassOf(<http://identifiers.org/hgnc/8907> <http://purl.obolibrary.org/obo/SO
 AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/8910> "PGR")
 SubClassOf(<http://identifiers.org/hgnc/8910> <http://purl.obolibrary.org/obo/SO_0000704>)
 
-# Class: <http://identifiers.org/hgnc/8912> (PHB)
+# Class: <http://identifiers.org/hgnc/8912> (PHB1)
 
-AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/8912> "PHB")
+AnnotationAssertion(rdfs:label <http://identifiers.org/hgnc/8912> "PHB1")
 SubClassOf(<http://identifiers.org/hgnc/8912> <http://purl.obolibrary.org/obo/SO_0000704>)
 
 # Class: <http://identifiers.org/hgnc/8918> (PHEX)

diff --git a/src/sparql/construct/construct-hgnc.sparql b/src/sparql/construct/construct-hgnc.sparql
@@ -6,14 +6,16 @@ prefix dc: <http://purl.org/dc/elements/1.1/>
 CONSTRUCT {
   ?gene a owl:Class ;
      rdfs:subClassOf <http://purl.obolibrary.org/obo/SO_0000704> ;
-     rdfs:label ?label ;
-     <http://purl.org/dc/elements/1.1/description> ?description . 
+     rdfs:label ?symbol ;
+     <http://purl.org/dc/terms/description> ?description . 
 }
 WHERE {
- ?gene rdfs:label ?label .
+ ?gene <https://w3id.org/biolink/vocab/category> <https://w3id.org/biolink/vocab/Gene> ;
+     <https://w3id.org/biolink/vocab/in_taxon> ?taxon_xref ;
+     <https://w3id.org/biolink/vocab/symbol> ?symbol .
 
  OPTIONAL {
-  ?gene <http://purl.org/dc/elements/1.1/description> ?description .
+  ?gene <http://purl.org/dc/terms/description> ?description .
  }
 
  FILTER (!isBlank(?gene))

diff --git a/src/sparql/construct/construct-ncbigene.sparql b/src/sparql/construct/construct-ncbigene.sparql
@@ -0,0 +1,29 @@
+prefix xsd: <http://www.w3.org/2001/XMLSchema#>
+prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+prefix owl: <http://www.w3.org/2002/07/owl#>
+prefix dc: <http://purl.org/dc/elements/1.1/>
+
+CONSTRUCT {
+  ?gene a owl:Class ;
+     rdfs:subClassOf <http://purl.obolibrary.org/obo/SO_0000704> ;
+     rdfs:label ?symbol ;
+     rdfs:subClassOf [
+        owl:onProperty <http://purl.obolibrary.org/obo/RO_0002162> ;
+        owl:someValuesFrom ?taxon ] ;
+     <http://purl.org/dc/terms/description> ?description . 
+}
+WHERE {
+ ?gene <https://w3id.org/biolink/vocab/category> <https://w3id.org/biolink/vocab/Gene> ;
+     <https://w3id.org/biolink/vocab/in_taxon> ?taxon_xref ;
+     <https://w3id.org/biolink/vocab/symbol> ?symbol .
+
+ OPTIONAL {
+  ?gene <http://purl.org/dc/terms/description> ?description .
+ }
+
+ FILTER (!isBlank(?gene))
+ FILTER ( regex(str(?gene), "^http://identifiers.org/ncbigene/"))
+ BIND(IRI(REPLACE(str(?taxon_xref), "NCBITaxon:","http://purl.obolibrary.org/obo/NCBITaxon_")) AS ?taxon)
+}
+
+
diff --git a/src/sparql/construct/construct-rare-characteristic-axioms.sparql b/src/sparql/construct/construct-rare-characteristic-axioms.sparql
@@ -6,7 +6,7 @@ prefix owl: <http://www.w3.org/2002/07/owl#>
 PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
 
 CONSTRUCT {
-  ?cls rdfs:subClassOf+ [
+  ?cls rdfs:subClassOf [
         owl:onProperty <http://purl.obolibrary.org/obo/RO_0000053> ;
         owl:someValuesFrom <http://purl.obolibrary.org/obo/MONDO_0021136> ] .
 }