Output interacting taxon in correct column.

geneontology · Jul 29, 2024 · 35b7fe2 · 35b7fe2
1 parent be87c34
commit 35b7fe2
Show file tree

Hide file tree

Showing 4 changed files with 8,613 additions and 3 deletions.
diff --git a/minerva-converter/src/main/java/org/geneontology/minerva/legacy/sparql/GPADRenderer.java b/minerva-converter/src/main/java/org/geneontology/minerva/legacy/sparql/GPADRenderer.java
@@ -46,7 +46,7 @@ public String render(GPADData data) {
             columns.add(data.getReference());
             columns.add(curieHandler.getCuri(data.getEvidence()));
             columns.add(data.getWithOrFrom().orElse(""));
-            columns.add(""); // not using interacting taxon in LEGO models
+            columns.add(data.getInteractingTaxonID().map(curieHandler::getCuri).orElse(""));
             columns.add(formatDate(data.getModificationDate()));
             columns.add(data.getAssignedBy());
             columns.add(formatAnnotationExtensions(data.getAnnotationExtensions()));

diff --git a/minerva-converter/src/main/java/org/geneontology/minerva/legacy/sparql/GPADSPARQLExport.java b/minerva-converter/src/main/java/org/geneontology/minerva/legacy/sparql/GPADSPARQLExport.java
@@ -43,11 +43,15 @@ public class GPADSPARQLExport {
     private static final String BP = "http://purl.obolibrary.org/obo/GO_0008150";
     private static final String CC = "http://purl.obolibrary.org/obo/GO_0005575";
     private static final Set<String> rootTerms = new HashSet<>(Arrays.asList(MF, BP, CC));
+
+    private static final String HAS_INPUT = "http://purl.obolibrary.org/obo/RO_0002233";
     private static final String ENABLES = "http://purl.obolibrary.org/obo/RO_0002327";
     private static final String CONTRIBUTES_TO = "http://purl.obolibrary.org/obo/RO_0002326";
     private static final Set<String> functionRelations = new HashSet<>(Arrays.asList(ENABLES, CONTRIBUTES_TO));
     private static final String EMAPA_NAMESPACE = "http://purl.obolibrary.org/obo/EMAPA_";
     private static final String UBERON_NAMESPACE = "http://purl.obolibrary.org/obo/UBERON_";
+
+    private static final String TAXON_NAMESPACE = "http://purl.obolibrary.org/obo/NCBITaxon_";
     private static final String inconsistentQuery =
             "PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>" +
                     "PREFIX owl: <http://www.w3.org/2002/07/owl#>" +
@@ -167,13 +171,18 @@ public Set<GPADData> getGPAD(WorkingMemory wm, IRI modelIRI) throws Inconsistent
                     annotationEvidences.forEach(currentEvidence -> {
                         String reference = currentEvidence.getReference();
                         Set<ConjunctiveExpression> goodExtensions = new HashSet<>();
+                        Optional<IRI> interactingTaxon = Optional.empty();
                         for (AnnotationExtension extension : possibleExtensions) {
                             if (extension.getTriple().getSubject().equals(annotation.getOntologyClassNode()) && !(extension.getTriple().getObject().equals(annotation.getObjectNode()))) {
                                 for (Explanation expl : allExplanations.get(extension.getTriple())) {
                                     boolean allFactsOfExplanationHaveRefMatchingAnnotation = toJava(expl.facts()).stream().map(fact -> allEvidences.getOrDefault(Bridge.jenaFromTriple(fact), Collections.emptySet())).allMatch(evidenceSet ->
                                             evidenceSet.stream().anyMatch(ev -> ev.getReference().equals(reference)));
                                     if (allFactsOfExplanationHaveRefMatchingAnnotation) {
-                                        goodExtensions.add(new DefaultConjunctiveExpression(IRI.create(extension.getTriple().getPredicate().getURI()), extension.getValueType()));
+                                        if (!interactingTaxon.isPresent() && extension.getTriple().getPredicate().getURI().equals(HAS_INPUT) && extension.getValueType().toString().startsWith(TAXON_NAMESPACE)) {
+                                            interactingTaxon = Optional.of(extension.getValueType());
+                                        } else {
+                                            goodExtensions.add(new DefaultConjunctiveExpression(IRI.create(extension.getTriple().getPredicate().getURI()), extension.getValueType()));
+                                        }
                                     }
                                 }
                             }
@@ -191,7 +200,7 @@ public Set<GPADData> getGPAD(WorkingMemory wm, IRI modelIRI) throws Inconsistent
                         final boolean rootMFWithOtherMF = annotation.getOntologyClass().toString().equals(MF) && gpsWithAnyMFNotRootMF.contains(annotation.getObject());
                         if (!rootViolation && !rootMFWithOtherMF) {
                             DefaultGPADData defaultGPADData = new DefaultGPADData(annotation.getObject(), annotation.getQualifier(), annotation.getOntologyClass(), goodExtensions,
-                                    reference, currentEvidence.getEvidence(), currentEvidence.getWithOrFrom(), Optional.empty(), currentEvidence.getModificationDate(),
+                                    reference, currentEvidence.getEvidence(), currentEvidence.getWithOrFrom(), interactingTaxon, currentEvidence.getModificationDate(),
                                     currentEvidence.getAssignedBy(), currentEvidence.getAnnotations());
                             defaultGPADData.setOperator(annotation.getOperator());
                             annotations.add(defaultGPADData);

diff --git a/minerva-converter/src/test/java/org/geneontology/minerva/legacy/sparql/GPADSPARQLTest.java b/minerva-converter/src/test/java/org/geneontology/minerva/legacy/sparql/GPADSPARQLTest.java
@@ -180,4 +180,21 @@ public void testFilterAnnotationsToRegulatedProcess() throws Exception {
         Assert.assertTrue(annotations.stream().noneMatch(a -> a.getObject().equals(gene) && a.getOntologyClass().equals(regulated)));
     }
 
+    @Test
+    public void testInteractingTaxon() {
+        Model model = ModelFactory.createDefaultModel();
+        model.read(this.getClass().getResourceAsStream("/MGI_MGI_2429397.ttl"), "", "ttl");
+        Set<Triple> triples = model.listStatements().toList().stream().map(s -> Bridge.tripleFromJena(s.asTriple())).collect(Collectors.toSet());
+        WorkingMemory mem = arachne.processTriples(JavaConverters.asScalaSetConverter(triples).asScala());
+        Set<GPADData> annotations = exporter.getGPAD(mem, IRI.create("http://test.org"));
+        IRI gene = IRI.create("http://identifiers.org/mgi/MGI:2429397");
+        IRI interactingTaxon = IRI.create("http://purl.obolibrary.org/obo/NCBITaxon_196620");
+        Assert.assertTrue(annotations.stream().anyMatch(
+                a -> a.getObject().equals(gene) &&
+                        a.getInteractingTaxonID().isPresent() &&
+                        a.getInteractingTaxonID().get().equals(interactingTaxon)));
+        Assert.assertTrue(annotations.stream().noneMatch(a ->
+                a.getAnnotationExtensions().stream().anyMatch(ce -> ce.getFiller().equals(interactingTaxon))));
+    }
+
 }