Update ontology term (ga4gh#1523)

* Add new simplified ontology term

* Refactor usage of ontology term

* Fix g2p errors

* Use updated client

* Update build test data

* Remove species fields

* Remove remaining sourcename from tests

docs/datarepo.rst

@@ -152,7 +152,7 @@ using command line options.
 
     $ ga4gh_repo add-referenceset registry.db hs37d5.fa.gz \
         --description "NCBI37 assembly of the human genome" \
-        --species '{"id": "9606", "term": "Homo sapiens", "sourceName": "NCBI", "sourceVersion: "1.0"}' \
+        --species '{"termId": "NCBI:9606", "term": "Homo sapiens"}' \
         --name NCBI37 \
         --sourceUri ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/reference/phase2_reference_assembly_sequence/hs37d5.fa.gz
 

ga4gh/server/datamodel/genotype_phenotype.py

@@ -212,14 +212,12 @@ def _formatOntologyTermObject(self, terms, element_type):
             terms = [terms]
         elements = []
         for term in terms:
-            if not issubclass(term.__class__, dict):
-                term = protocol.toJsonDict(term)
-            if term['id']:
+            if term.term_id:
                 elements.append('?{} = <{}> '.format(
-                    element_type, term['id']))
+                    element_type, term.term_id))
             else:
                 elements.append('?{} = <{}> '.format(
-                    element_type, self._toNamespaceURL(term['term'])))
+                    element_type, self._toNamespaceURL(term.term)))
         elementClause = "({})".format(" || ".join(elements))
         return elementClause
 
@@ -403,10 +401,7 @@ def _toGA4GH(self, association, featureSets=[]):
 
         term = protocol.OntologyTerm()
         term.term = association['evidence_type']
-        term.id = phenotype['id']
-        term.source_version = self._version
-        term.source_name = self._getPrefix(
-            self._getPrefixURL(association['id']))
+        term.term_id = phenotype['id']
         evidence.evidence_type.MergeFrom(term)
 
         evidence.description = self._getIdentifier(association['evidence'])
@@ -424,10 +419,7 @@ def _toGA4GH(self, association, featureSets=[]):
 
         term = protocol.OntologyTerm()
         term.term = environment['id']
-        term.id = 'http://purl.obolibrary.org/obo/RO_0002606'
-        term.source_version = self._version
-        term.source_name = self._getPrefix(
-            self._getPrefixURL(association['id']))
+        term.term_id = 'http://purl.obolibrary.org/obo/RO_0002606'
         environmentalContext.environment_type.MergeFrom(term)
 
         fpa.environmental_contexts.extend([environmentalContext])
@@ -436,10 +428,7 @@ def _toGA4GH(self, association, featureSets=[]):
         phenotypeInstance = protocol.PhenotypeInstance()
         term = protocol.OntologyTerm()
         term.term = phenotype[TYPE]
-        term.id = phenotype['id']
-        term.source_version = self._version
-        term.source_name = self._getPrefix(
-            self._getPrefixURL(association['id']))
+        term.term_id = phenotype['id']
         phenotypeInstance.type.MergeFrom(term)
 
         phenotypeInstance.description = phenotype[LABEL]

ga4gh/server/datamodel/genotype_phenotype_featureset.py

@@ -110,7 +110,7 @@ def _getFeatureById(self, featureId):
         for featureType in sorted(feature[TYPE]):
             if "obolibrary" in featureType:
                 term.term = self._featureTypeLabel(featureType)
-                term.id = featureType
+                term.term_id = featureType
                 pbFeature.feature_type.MergeFrom(term)
                 break
 

ga4gh/server/datamodel/ontologies.py

@@ -12,8 +12,6 @@
 import ga4gh.server.exceptions as exceptions
 import ga4gh.server.datamodel.obo_parser as obo_parser
 
-import ga4gh.schemas.pb as pb
-
 
 SEQUENCE_ONTOLOGY_PREFIX = "SO"
 
@@ -137,7 +135,5 @@ def getGaTermByName(self, name):
             termId = termIds[0]
         term = protocol.OntologyTerm()
         term.term = name
-        term.id = termId
-        term.source_name = self._sourceName
-        term.source_version = pb.string(self._sourceVersion)
+        term.term_id = termId
         return term

ga4gh/server/datamodel/references.py

@@ -211,10 +211,8 @@ def toProtocolElement(self):
         if self.getSpecies():
             term = protocol.fromJson(
                 json.dumps(self.getSpecies()), protocol.OntologyTerm)
-            ret.species.id = term.id
+            ret.species.term_id = term.term_id
             ret.species.term = term.term
-            ret.species.source_name = term.source_name
-            ret.species.source_version = term.source_version
         ret.source_accessions.extend(self.getSourceAccessions())
         ret.source_uri = pb.string(self.getSourceUri())
         ret.name = self.getLocalId()
@@ -355,10 +353,8 @@ def toProtocolElement(self):
         if self.getSpecies():
             term = protocol.fromJson(
                 json.dumps(self.getSpecies()), protocol.OntologyTerm)
-            reference.species.id = term.id
+            reference.species.term_id = term.term_id
             reference.species.term = term.term
-            reference.species.source_name = term.source_name
-            reference.species.source_version = term.source_version
         reference.source_accessions.extend(self.getSourceAccessions())
         reference.source_divergence = pb.int(self.getSourceDivergence())
         reference.source_uri = self.getSourceUri()
@@ -403,8 +399,7 @@ def __init__(self, localId, randomSeed=0, numReferences=1):
         self._assemblyId = str(random.randint(0, 2**32))
         self._isDerived = bool(random.randint(0, 1))
         self._species = json.loads(
-                    '{"sourceName": "NCBI", "sourceVersion": "",'
-                    + '"term": "Homo sapiens", "id": "9606"}')
+                    '{"term": "Homo sapiens", "termId": "9606"}')
         self._sourceAccessions = []
         for i in range(random.randint(1, 3)):
                 self._sourceAccessions.append("sim_accession_{}".format(
@@ -437,8 +432,7 @@ def __init__(self, parentContainer, localId, randomSeed=0, length=200):
         if self._isDerived:
             self._sourceDivergence = rng.uniform(0, 0.1)
         self._species = json.loads(
-                            '{"sourceName": "NCBI", "sourceVersion": "",'
-                            + '"term": "Homo sapiens", "id": "9606"}')
+                            '{"term": "Homo sapiens", "termId": "9606"}')
         self._sourceAccessions = []
         for i in range(random.randint(1, 3)):
                 self._sourceAccessions.append("sim_accession_{}".format(

ga4gh/server/datamodel/sequence_annotations.py

@@ -228,9 +228,7 @@ def _getRandomfeatureType(self, randomNumberGenerator):
             ("exon", "SO:0000147")]
         term = protocol.OntologyTerm()
         ontologyTuple = randomNumberGenerator.choice(ontologyTuples)
-        term.term, term.id = ontologyTuple[0], ontologyTuple[1]
-        term.source_name = "sequenceOntology"
-        term.source_version = "0"
+        term.term, term.term_id = ontologyTuple[0], ontologyTuple[1]
         return term
 
     def _generateSimulatedFeature(self, randomNumberGenerator):

ga4gh/server/datamodel/variants.py

@@ -993,9 +993,7 @@ def _getRandomOntologyTerm(self, randomNumberGenerator):
             ("exon_variant", "SO:0001791")]
         term = protocol.OntologyTerm()
         ontologyTuple = randomNumberGenerator.choice(ontologyTuples)
-        term.term, term.id = ontologyTuple[0], ontologyTuple[1]
-        term.source_name = "ontology"
-        term.source_version = "0"
+        term.term, term.term_id = ontologyTuple[0], ontologyTuple[1]
         return term
 
     def _addTranscriptEffectOntologyTerm(self, effect, randomNumberGenerator):

ga4gh/server/paging.py

@@ -274,10 +274,10 @@ def _checkIdEquality(self, requestedEffect, effect):
         present in an annotation are equal.
         """
         return self._idPresent(requestedEffect) and (
-            effect.id == requestedEffect.id)
+            effect.term_id == requestedEffect.term_id)
 
     def _idPresent(self, requestedEffect):
-        return requestedEffect.id != ""
+        return requestedEffect.term_id != ""
 
     def _matchAnyEffects(self, effect):
         ret = False

scripts/build_test_data.py

@@ -37,8 +37,8 @@ def buildTestData(
     pattern = os.path.join(prefix, "referenceSets", "*.fa.gz")
     for dataFile in glob.glob(pattern):
         run("add-referenceset", repoFile, useRelativePath, dataFile,
-            "--species ", '\'{"id": "9606", "term": "Homo sapiens", '
-                          '"sourceName": "NCBI", "sourceVersion": "1.0"}\'')
+            "--species ", '\'{"termId": "NCBI:9606", '
+                          '"term": "Homo sapiens"}\'')
 
     pattern = os.path.join(prefix, "ontologies", "*.obo")
     for dataFile in glob.glob(pattern):

tests/datadriven/test_ontologies.py

@@ -53,10 +53,7 @@ def testGoodMappings(self):
             self.assertTrue(protocol.validate(protocol.toJson(gaTerm),
                                               OntologyTerm))
             self.assertEqual(gaTerm.term, term.name)
-            self.assertIn(gaTerm.id, ontology.getTermIds(term.name))
-            self.assertEqual(
-                gaTerm.source_version, ontology.getSourceVersion())
-            self.assertEqual(gaTerm.source_name, ontology.getName())
+            self.assertIn(gaTerm.term_id, ontology.getTermIds(term.name))
 
     def testBadMappings(self):
         for badName in ["Not a term", None, 1234]:

tests/end_to_end/test_g2p.py

@@ -199,7 +199,7 @@ def testFeaturesSearchById(self):
         self.assertIsNotNone(feature)
         self.assertEqual(request.feature_id, feature.id)
         self.assertIsNotNone(feature.feature_type)
-        self.assertIsNotNone(feature.feature_type.id)
+        self.assertIsNotNone(feature.feature_type.term_id)
         self.assertEqual(feature.reference_name,  "chr10")
         self.assertEqual(feature.start,  43617416)
         self.assertEqual(feature.end,  43617416)
@@ -258,7 +258,7 @@ def testPhenotypesSearchOntologyTerm(self):
         request = protocol.SearchPhenotypesRequest()
         request.phenotype_association_set_id = \
             self.getPhenotypeAssociationSetId()
-        request.type.id = "http://ohsu.edu/cgd/5c895709"
+        request.type.term_id = "http://ohsu.edu/cgd/5c895709"
         postUrl = '/phenotypes/search'
         response = self.sendSearchRequest(
             postUrl,
@@ -271,7 +271,7 @@ def testPhenotypeSearchQualifiersSensitivity(self):
         request.phenotype_association_set_id = \
             self.getPhenotypeAssociationSetId()
         ontologyterm = protocol.OntologyTerm()
-        ontologyterm.id = "http://ohsu.edu/cgd/sensitivity"
+        ontologyterm.term_id = "http://ohsu.edu/cgd/sensitivity"
         request.qualifiers.extend([ontologyterm])
         postUrl = '/phenotypes/search'
         response = self.sendSearchRequest(
@@ -285,7 +285,7 @@ def testPhenotypeSearchQualifiersSensitivityPATO_0000396(self):
         request.phenotype_association_set_id = \
             self.getPhenotypeAssociationSetId()
         ontologyterm = protocol.OntologyTerm()
-        ontologyterm.id = "http://purl.obolibrary.org/obo/PATO_0000396"
+        ontologyterm.term_id = "http://purl.obolibrary.org/obo/PATO_0000396"
         request.qualifiers.extend([ontologyterm])
         postUrl = '/phenotypes/search'
         response = self.sendSearchRequest(
@@ -299,9 +299,9 @@ def testPhenotypeSearchMultipleQualifiers(self):
         request.phenotype_association_set_id = \
             self.getPhenotypeAssociationSetId()
         ontologyterm = protocol.OntologyTerm()
-        ontologyterm.id = "http://purl.obolibrary.org/obo/PATO_0000396"
+        ontologyterm.term_id = "http://purl.obolibrary.org/obo/PATO_0000396"
         ontologyterm2 = protocol.OntologyTerm()
-        ontologyterm2.id = "http://purl.obolibrary.org/obo/PATO_0000460"
+        ontologyterm2.term_id = "http://purl.obolibrary.org/obo/PATO_0000460"
         request.qualifiers.extend([ontologyterm, ontologyterm2])
         postUrl = '/phenotypes/search'
         response = self.sendSearchRequest(
@@ -341,7 +341,8 @@ def testPhenotypesSearchMultipleTerms(self):
         request.phenotype_association_set_id = \
             self.getPhenotypeAssociationSetId()
         request.description = "Melanoma, NOS with response to therapy"
-        request.age_of_onset.id = "http://purl.obolibrary.org/obo/HP_0003581"
+        request.age_of_onset.term_id = \
+            "http://purl.obolibrary.org/obo/HP_0003581"
         postUrl = '/phenotypes/search'
         response = self.sendSearchRequest(
             postUrl,

tests/unit/test_bio_metadata.py

@@ -12,8 +12,6 @@
 import ga4gh.server.datamodel.bio_metadata as bioMetadata
 import ga4gh.server.protocol as protocol
 
-import ga4gh.schemas.pb as pb
-
 
 class TestIndividuals(unittest.TestCase):
     """
@@ -23,9 +21,7 @@ def testToProtocolElement(self):
         dataset = datasets.Dataset('dataset1')
         term = protocol.OntologyTerm()
         term.term = "male genotypic sex"
-        term.id = "PATO:0020001"
-        term.source_name = "PATO"
-        term.source_version = pb.string("2015-11-18")
+        term.term_id = "PATO:0020001"
         # Write out a valid input
         print(protocol.toJsonDict(term))
         validIndividual = protocol.Individual(

tests/unit/test_simulated_stack.py

@@ -165,13 +165,9 @@ def verifyReferenceSetsEqual(self, gaReferenceSet, referenceSet):
         sp = protocol.fromJson(
                 json.dumps(referenceSet.getSpecies()), protocol.OntologyTerm)
         self.assertEqual(
-            gaReferenceSet.species.id, sp.id)
+            gaReferenceSet.species.term_id, sp.term_id)
         self.assertEqual(
             gaReferenceSet.species.term, sp.term)
-        self.assertEqual(
-            gaReferenceSet.species.source_name, sp.source_name)
-        self.assertEqual(
-            gaReferenceSet.species.source_version, sp.source_version)
         self.assertEqual(
             gaReferenceSet.assembly_id, referenceSet.getAssemblyId())
         self.assertEqual(
@@ -204,14 +200,8 @@ def verifyReferencesEqual(self, gaReference, reference):
         self.assertEqual(gaReference.md5checksum, reference.getMd5Checksum())
         sp = protocol.fromJson(
                 json.dumps(reference.getSpecies()), protocol.OntologyTerm)
-        self.assertEqual(gaReference.species.id, sp.id)
+        self.assertEqual(gaReference.species.term_id, sp.term_id)
         self.assertEqual(gaReference.species.term, sp.term)
-        self.assertEqual(
-            gaReference.species.source_name,
-            sp.source_name)
-        self.assertEqual(
-            gaReference.species.source_version,
-            sp.source_version)
         self.assertEqual(gaReference.source_uri, reference.getSourceUri())
         self.assertEqual(
             gaReference.source_accessions, reference.getSourceAccessions())
@@ -651,7 +641,7 @@ def testVariantAnnotationsSearch(self):
         request.end = 10
         request.reference_name = "1"
 
-        request.effects.add().id = "ThisIsNotAnEffect"
+        request.effects.add().term_id = "ThisIsNotAnEffect"
 
         response = self.sendJsonPostRequest(path, protocol.toJson(request))
         responseData = protocol.fromJson(response.data, protocol.
@@ -680,8 +670,8 @@ def testVariantAnnotationsSearch(self):
         request.start = 0
         request.end = 5
         request.reference_name = "1"
-        request.effects.add().id = "SO:0001627"
-        request.effects.add().id = "B4DID"
+        request.effects.add().term_id = "SO:0001627"
+        request.effects.add().term_id = "B4DID"
         response = self.sendJsonPostRequest(path, protocol.toJson(request))
         responseData = protocol.fromJson(response.data, protocol.
                                          SearchVariantAnnotationsResponse)
@@ -693,8 +683,8 @@ def testVariantAnnotationsSearch(self):
             effectPresent = False
             for effect in ann.transcript_effects:
                 for featureType in effect.effects:
-                    if featureType.id in map(
-                            lambda e: e.id, request.effects):
+                    if featureType.term_id in map(
+                            lambda e: e.term_id, request.effects):
                         effectPresent = True
             self.assertEquals(
                 True, effectPresent,
@@ -705,8 +695,8 @@ def testVariantAnnotationsSearch(self):
         request.start = 0
         request.end = 5
         request.reference_name = "1"
-        request.effects.add().id = "B4DID"
-        request.effects.add().id = "SO:0001627"
+        request.effects.add().term_id = "B4DID"
+        request.effects.add().term_id = "SO:0001627"
         response = self.sendJsonPostRequest(path, protocol.toJson(request))
         responseData = protocol.fromJson(response.data, protocol.
                                          SearchVariantAnnotationsResponse)
@@ -718,8 +708,8 @@ def testVariantAnnotationsSearch(self):
             effectPresent = False
             for effect in ann.transcript_effects:
                 for featureType in effect.effects:
-                    if featureType.id in map(
-                            lambda e: e.id, request.effects):
+                    if featureType.term_id in map(
+                            lambda e: e.term_id, request.effects):
                         effectPresent = True
             self.assertEquals(
                 True,
@@ -731,7 +721,7 @@ def testVariantAnnotationsSearch(self):
         request.start = 0
         request.end = 5
         request.reference_name = "1"
-        request.effects.add().id = "SO:0001627"
+        request.effects.add().term_id = "SO:0001627"
         response = self.sendJsonPostRequest(path, protocol.toJson(request))
         responseData = protocol.fromJson(response.data, protocol.
                                          SearchVariantAnnotationsResponse)
@@ -744,8 +734,8 @@ def testVariantAnnotationsSearch(self):
                              "Transcript effects should be unique")
             for effect in ann.transcript_effects:
                 for featureType in effect.effects:
-                    if featureType.id in map(
-                            lambda e: e.id, request.effects):
+                    if featureType.term_id in map(
+                            lambda e: e.term_id, request.effects):
                         effectPresent = True
             self.assertEquals(True, effectPresent,
                               "The ontology term should appear at least once")
@@ -755,8 +745,8 @@ def testVariantAnnotationsSearch(self):
         request.start = 0
         request.end = 10
         request.reference_name = "1"
-        request.effects.add().id = "SO:0001627"
-        request.effects.add().id = "SO:0001791"
+        request.effects.add().term_id = "SO:0001627"
+        request.effects.add().term_id = "SO:0001791"
         response = self.sendJsonPostRequest(path, protocol.toJson(request))
         responseData = protocol.fromJson(response.data, protocol.
                                          SearchVariantAnnotationsResponse)