genotype and variant support (#740)

Adds initial support for genotypes & variants and their associations to the API & UI --------- Co-authored-by: glass-ships <glass.ships@outlook.com>
monarch-initiative · Jun 11, 2024 · c106f34 · c106f34
1 parent a632971
commit c106f34
Show file tree

Hide file tree

Showing 41 changed files with 98,973 additions and 102,322 deletions.
diff --git a/backend/src/monarch_py/datamodels/category_enums.py b/backend/src/monarch_py/datamodels/category_enums.py
@@ -4,13 +4,15 @@
 class EntityCategory(Enum):
     """Entity categories"""
 
+    SEQUENCE_VARIANT = "biolink:SequenceVariant"
     GENE = "biolink:Gene"
+    GENOTYPE = "biolink:Genotype"
     PHENOTYPIC_FEATURE = "biolink:PhenotypicFeature"
     BIOLOGICAL_PROCESS_OR_ACTIVITY = "biolink:BiologicalProcessOrActivity"
-    GROSS_ANATOMICAL_STRUCTURE = "biolink:GrossAnatomicalStructure"
     DISEASE = "biolink:Disease"
-    PATHWAY = "biolink:Pathway"
+    GROSS_ANATOMICAL_STRUCTURE = "biolink:GrossAnatomicalStructure"
     CELL = "biolink:Cell"
+    PATHWAY = "biolink:Pathway"
     NAMED_THING = "biolink:NamedThing"
     ANATOMICAL_ENTITY = "biolink:AnatomicalEntity"
     CELLULAR_COMPONENT = "biolink:CellularComponent"
@@ -20,77 +22,67 @@ class EntityCategory(Enum):
     MOLECULAR_ACTIVITY = "biolink:MolecularActivity"
     PROTEIN = "biolink:Protein"
     CELLULAR_ORGANISM = "biolink:CellularOrganism"
-    PHENOTYPIC_QUALITY = "biolink:PhenotypicQuality"
     VERTEBRATE = "biolink:Vertebrate"
     VIRUS = "biolink:Virus"
     BEHAVIORAL_FEATURE = "biolink:BehavioralFeature"
+    CHEMICAL_ENTITY = "biolink:ChemicalEntity"
     LIFE_STAGE = "biolink:LifeStage"
     PATHOLOGICAL_PROCESS = "biolink:PathologicalProcess"
-    CHEMICAL_ENTITY = "biolink:ChemicalEntity"
     DRUG = "biolink:Drug"
-    ORGANISM_TAXON = "biolink:OrganismTaxon"
-    SEQUENCE_VARIANT = "biolink:SequenceVariant"
     SMALL_MOLECULE = "biolink:SmallMolecule"
+    ORGANISM_TAXON = "biolink:OrganismTaxon"
     INFORMATION_CONTENT_ENTITY = "biolink:InformationContentEntity"
     NUCLEIC_ACID_ENTITY = "biolink:NucleicAcidEntity"
     EVIDENCE_TYPE = "biolink:EvidenceType"
-    GEOGRAPHIC_EXPOSURE = "biolink:GeographicExposure"
     RNAPRODUCT = "biolink:RNAProduct"
     TRANSCRIPT = "biolink:Transcript"
     FUNGUS = "biolink:Fungus"
     PLANT = "biolink:Plant"
-    DATASET = "biolink:Dataset"
-    INVERTEBRATE = "biolink:Invertebrate"
-    POPULATION_OF_INDIVIDUAL_ORGANISMS = "biolink:PopulationOfIndividualOrganisms"
-    PROTEIN_FAMILY = "biolink:ProteinFamily"
+    PROCESSED_MATERIAL = "biolink:ProcessedMaterial"
     ACTIVITY = "biolink:Activity"
     AGENT = "biolink:Agent"
-    CHEMICAL_EXPOSURE = "biolink:ChemicalExposure"
     CONFIDENCE_LEVEL = "biolink:ConfidenceLevel"
+    DATASET = "biolink:Dataset"
     ENVIRONMENTAL_FEATURE = "biolink:EnvironmentalFeature"
-    EXON = "biolink:Exon"
     GENETIC_INHERITANCE = "biolink:GeneticInheritance"
-    GENOME = "biolink:Genome"
-    GENOTYPE = "biolink:Genotype"
     HAPLOTYPE = "biolink:Haplotype"
-    HUMAN = "biolink:Human"
-    INDIVIDUAL_ORGANISM = "biolink:IndividualOrganism"
+    INVERTEBRATE = "biolink:Invertebrate"
     MAMMAL = "biolink:Mammal"
-    MATERIAL_SAMPLE = "biolink:MaterialSample"
-    MICRO_RNA = "biolink:MicroRNA"
-    PATENT = "biolink:Patent"
-    PROTEIN_DOMAIN = "biolink:ProteinDomain"
+    POPULATION_OF_INDIVIDUAL_ORGANISMS = "biolink:PopulationOfIndividualOrganisms"
+    PROTEIN_FAMILY = "biolink:ProteinFamily"
     PUBLICATION = "biolink:Publication"
-    REGULATORY_REGION = "biolink:RegulatoryRegion"
-    STUDY = "biolink:Study"
-    TREATMENT = "biolink:Treatment"
-    WEB_PAGE = "biolink:WebPage"
     ACCESSIBLE_DNA_REGION = "biolink:AccessibleDnaRegion"
-    ARTICLE = "biolink:Article"
-    ATTRIBUTE = "biolink:Attribute"
     BACTERIUM = "biolink:Bacterium"
-    BIOLOGICAL_ENTITY = "biolink:BiologicalEntity"
     BIOLOGICAL_SEX = "biolink:BiologicalSex"
     CELL_LINE = "biolink:CellLine"
+    CHEMICAL_EXPOSURE = "biolink:ChemicalExposure"
     CHEMICAL_MIXTURE = "biolink:ChemicalMixture"
-    CODING_SEQUENCE = "biolink:CodingSequence"
     DATASET_DISTRIBUTION = "biolink:DatasetDistribution"
     DIAGNOSTIC_AID = "biolink:DiagnosticAid"
     DRUG_EXPOSURE = "biolink:DrugExposure"
     ENVIRONMENTAL_PROCESS = "biolink:EnvironmentalProcess"
     EVENT = "biolink:Event"
+    EXON = "biolink:Exon"
+    GENOME = "biolink:Genome"
     GENOTYPIC_SEX = "biolink:GenotypicSex"
-    NONCODING_RNAPRODUCT = "biolink:NoncodingRNAProduct"
+    HUMAN = "biolink:Human"
+    INDIVIDUAL_ORGANISM = "biolink:IndividualOrganism"
+    MATERIAL_SAMPLE = "biolink:MaterialSample"
+    MICRO_RNA = "biolink:MicroRNA"
     ORGANISMAL_ENTITY = "biolink:OrganismalEntity"
+    PATENT = "biolink:Patent"
     PHENOTYPIC_SEX = "biolink:PhenotypicSex"
     POLYPEPTIDE = "biolink:Polypeptide"
-    PROCEDURE = "biolink:Procedure"
-    PROCESSED_MATERIAL = "biolink:ProcessedMaterial"
+    PROTEIN_DOMAIN = "biolink:ProteinDomain"
     REAGENT_TARGETED_GENE = "biolink:ReagentTargetedGene"
+    REGULATORY_REGION = "biolink:RegulatoryRegion"
     SI_RNA = "biolink:SiRNA"
     SNV = "biolink:Snv"
+    STUDY = "biolink:Study"
     STUDY_VARIABLE = "biolink:StudyVariable"
     TRANSCRIPTION_FACTOR_BINDING_SITE = "biolink:TranscriptionFactorBindingSite"
+    TREATMENT = "biolink:Treatment"
+    WEB_PAGE = "biolink:WebPage"
     ZYGOSITY = "biolink:Zygosity"
 
 
@@ -99,6 +91,7 @@ class AssociationCategory(Enum):
 
     PAIRWISE_GENE_TO_GENE_INTERACTION = "biolink:PairwiseGeneToGeneInteraction"
     GENE_TO_EXPRESSION_SITE_ASSOCIATION = "biolink:GeneToExpressionSiteAssociation"
+    VARIANT_TO_GENE_ASSOCIATION = "biolink:VariantToGeneAssociation"
     MACROMOLECULAR_MACHINE_TO_BIOLOGICAL_PROCESS_ASSOCIATION = (
         "biolink:MacromolecularMachineToBiologicalProcessAssociation"
     )
@@ -111,14 +104,19 @@ class AssociationCategory(Enum):
     )
     ASSOCIATION = "biolink:Association"
     GENE_TO_GENE_HOMOLOGY_ASSOCIATION = "biolink:GeneToGeneHomologyAssociation"
+    GENOTYPE_TO_PHENOTYPIC_FEATURE_ASSOCIATION = "biolink:GenotypeToPhenotypicFeatureAssociation"
     DISEASE_TO_PHENOTYPIC_FEATURE_ASSOCIATION = "biolink:DiseaseToPhenotypicFeatureAssociation"
     GENE_TO_PATHWAY_ASSOCIATION = "biolink:GeneToPathwayAssociation"
+    DISEASE_OR_PHENOTYPIC_FEATURE_TO_LOCATION_ASSOCIATION = "biolink:DiseaseOrPhenotypicFeatureToLocationAssociation"
     CHEMICAL_TO_PATHWAY_ASSOCIATION = "biolink:ChemicalToPathwayAssociation"
+    GENOTYPE_TO_DISEASE_ASSOCIATION = "biolink:GenotypeToDiseaseAssociation"
     CORRELATED_GENE_TO_DISEASE_ASSOCIATION = "biolink:CorrelatedGeneToDiseaseAssociation"
     DISEASE_OR_PHENOTYPIC_FEATURE_TO_GENETIC_INHERITANCE_ASSOCIATION = (
         "biolink:DiseaseOrPhenotypicFeatureToGeneticInheritanceAssociation"
     )
+    VARIANT_TO_DISEASE_ASSOCIATION = "biolink:VariantToDiseaseAssociation"
     CAUSAL_GENE_TO_DISEASE_ASSOCIATION = "biolink:CausalGeneToDiseaseAssociation"
+    CHEMICAL_TO_DISEASE_OR_PHENOTYPIC_FEATURE_ASSOCIATION = "biolink:ChemicalToDiseaseOrPhenotypicFeatureAssociation"
 
 
 class AssociationPredicate(Enum):
@@ -127,32 +125,33 @@ class AssociationPredicate(Enum):
     INTERACTS_WITH = "biolink:interacts_with"
     EXPRESSED_IN = "biolink:expressed_in"
     HAS_PHENOTYPE = "biolink:has_phenotype"
+    IS_SEQUENCE_VARIANT_OF = "biolink:is_sequence_variant_of"
     ENABLES = "biolink:enables"
     ACTIVELY_INVOLVED_IN = "biolink:actively_involved_in"
     ORTHOLOGOUS_TO = "biolink:orthologous_to"
     LOCATED_IN = "biolink:located_in"
     SUBCLASS_OF = "biolink:subclass_of"
+    RELATED_TO = "biolink:related_to"
     PARTICIPATES_IN = "biolink:participates_in"
     ACTS_UPSTREAM_OF_OR_WITHIN = "biolink:acts_upstream_of_or_within"
-    RELATED_TO = "biolink:related_to"
     ACTIVE_IN = "biolink:active_in"
     PART_OF = "biolink:part_of"
+    MODEL_OF = "biolink:model_of"
+    CAUSES = "biolink:causes"
     ACTS_UPSTREAM_OF = "biolink:acts_upstream_of"
     HAS_MODE_OF_INHERITANCE = "biolink:has_mode_of_inheritance"
-    GENE_ASSOCIATED_WITH_CONDITION = "biolink:gene_associated_with_condition"
     CONTRIBUTES_TO = "biolink:contributes_to"
-    CAUSES = "biolink:causes"
+    GENE_ASSOCIATED_WITH_CONDITION = "biolink:gene_associated_with_condition"
+    TREATS_OR_APPLIED_OR_STUDIED_TO_TREAT = "biolink:treats_or_applied_or_studied_to_treat"
     COLOCALIZES_WITH = "biolink:colocalizes_with"
-    ACTS_UPSTREAM_OF_OR_WITHIN_POSITIVE_EFFECT = "biolink:acts_upstream_of_or_within_positive_effect"
     ACTS_UPSTREAM_OF_POSITIVE_EFFECT = "biolink:acts_upstream_of_positive_effect"
-    ACTS_UPSTREAM_OF_OR_WITHIN_NEGATIVE_EFFECT = "biolink:acts_upstream_of_or_within_negative_effect"
+    ACTS_UPSTREAM_OF_OR_WITHIN_POSITIVE_EFFECT = "biolink:acts_upstream_of_or_within_positive_effect"
     ACTS_UPSTREAM_OF_NEGATIVE_EFFECT = "biolink:acts_upstream_of_negative_effect"
+    ACTS_UPSTREAM_OF_OR_WITHIN_NEGATIVE_EFFECT = "biolink:acts_upstream_of_or_within_negative_effect"
 
 
 class MappingPredicate(Enum):
     """Mapping predicates"""
 
     EXACT_MATCH = "skos:exactMatch"
-    CLOSE_MATCH = "skos:closeMatch"
     BROAD_MATCH = "skos:broadMatch"
-    NARROW_MATCH = "skos:narrowMatch"
diff --git a/backend/src/monarch_py/utils/association_type_mappings.yaml b/backend/src/monarch_py/utils/association_type_mappings.yaml
@@ -36,3 +36,15 @@
 - subject_label: Correlated Gene
   object_label: Correlated Disease
   category: "biolink:CorrelatedGeneToDiseaseAssociation"
+- subject_label: Variant to Gene
+  object_label: Variant to Gene
+  category: "biolink:VariantToGeneAssociation"
+- subject_label: Variant to Disease
+  object_label: Variant to Disease
+  category: "biolink:VariantToDiseaseAssociation"
+- subject_label: Genotype to Phenotype
+  object_label:  Genotype to Phenotype
+  category: "biolink:GenotypeToPhenotypicFeatureAssociation"
+- subject_label: Disease Model
+  object_label: Disease Model
+  category: "biolink:GenotypeToDiseaseAssociation"
diff --git a/backend/src/monarch_py/utils/association_type_utils.py b/backend/src/monarch_py/utils/association_type_utils.py
@@ -50,11 +50,9 @@ def get_association_type_mapping_by_query_string(
     Raises: ValueError if no match is found
     """
 
-    categories = parse_query_string_for_category(query_string)
+    category = parse_query_string_for_category(query_string)
 
-    matching_types = [
-        a_type for a_type in AssociationTypeMappings.get_mappings() if set(a_type.category) == set(categories)
-    ]
+    matching_types = [a_type for a_type in AssociationTypeMappings.get_mappings() if a_type.category == category]
 
     if len(matching_types) == 0:
         raise ValueError(f"No matching association type found for query string: [{query_string}]")

diff --git a/backend/src/monarch_py/utils/utils.py b/backend/src/monarch_py/utils/utils.py
@@ -68,7 +68,7 @@ def set_log_level(log_level: str):
 
 def get_links_for_field(field: List[str]) -> List[ExpandedCurie]:
     # TODO should be able to remove curie.replace("PMID", "PUBMED")) since the converter should handle prefix synonyms
-    expanded_curies = [ExpandedCurie(id=curie, url=get_link_for_curie(curie)) for curie in field]
+    expanded_curies = [ExpandedCurie(id=curie, url=get_link_for_curie(curie)) for curie in field if ":" in curie]
     return expanded_curies
 
 

diff --git a/backend/tests/fixtures/association_counts.py b/backend/tests/fixtures/association_counts.py
@@ -7,10 +7,12 @@ def association_counts():
         "items": [
             {
                 "label": "Phenotype to Disease",
-                "count": 3932,
+                "count": 3941,
                 "category": "biolink:DiseaseToPhenotypicFeatureAssociation",
             },
             {"label": "Causal Gene", "count": 126, "category": "biolink:CausalGeneToDiseaseAssociation"},
             {"label": "Correlated Gene", "count": 146, "category": "biolink:CorrelatedGeneToDiseaseAssociation"},
+            {"label": "Variant to Disease", "count": 1, "category": "biolink:VariantToDiseaseAssociation"},
+            {"label": "Disease Model", "count": 243, "category": "biolink:GenotypeToDiseaseAssociation"},
         ]
     }
diff --git a/backend/tests/fixtures/association_counts_query.py b/backend/tests/fixtures/association_counts_query.py
@@ -23,6 +23,10 @@ def association_counts_query():
             '(category:"biolink:MacromolecularMachineToBiologicalProcessAssociation") AND (subject:"MONDO:0020121" OR subject_closure:"MONDO:0020121")',
             '(category:"biolink:CausalGeneToDiseaseAssociation") AND (subject:"MONDO:0020121" OR subject_closure:"MONDO:0020121")',
             '(category:"biolink:CorrelatedGeneToDiseaseAssociation") AND (subject:"MONDO:0020121" OR subject_closure:"MONDO:0020121")',
+            '(category:"biolink:VariantToGeneAssociation") AND (subject:"MONDO:0020121" OR subject_closure:"MONDO:0020121")',
+            '(category:"biolink:VariantToDiseaseAssociation") AND (subject:"MONDO:0020121" OR subject_closure:"MONDO:0020121")',
+            '(category:"biolink:GenotypeToPhenotypicFeatureAssociation") AND (subject:"MONDO:0020121" OR subject_closure:"MONDO:0020121")',
+            '(category:"biolink:GenotypeToDiseaseAssociation") AND (subject:"MONDO:0020121" OR subject_closure:"MONDO:0020121")',
             '(category:"biolink:DiseaseToPhenotypicFeatureAssociation") AND (object:"MONDO:0020121" OR object_closure:"MONDO:0020121")',
             '(category:"biolink:GeneToPhenotypicFeatureAssociation") AND (object:"MONDO:0020121" OR object_closure:"MONDO:0020121")',
             '(category:"biolink:PairwiseGeneToGeneInteraction") AND (object:"MONDO:0020121" OR object_closure:"MONDO:0020121")',
@@ -35,6 +39,10 @@ def association_counts_query():
             '(category:"biolink:MacromolecularMachineToBiologicalProcessAssociation") AND (object:"MONDO:0020121" OR object_closure:"MONDO:0020121")',
             '(category:"biolink:CausalGeneToDiseaseAssociation") AND (object:"MONDO:0020121" OR object_closure:"MONDO:0020121")',
             '(category:"biolink:CorrelatedGeneToDiseaseAssociation") AND (object:"MONDO:0020121" OR object_closure:"MONDO:0020121")',
+            '(category:"biolink:VariantToGeneAssociation") AND (object:"MONDO:0020121" OR object_closure:"MONDO:0020121")',
+            '(category:"biolink:VariantToDiseaseAssociation") AND (object:"MONDO:0020121" OR object_closure:"MONDO:0020121")',
+            '(category:"biolink:GenotypeToPhenotypicFeatureAssociation") AND (object:"MONDO:0020121" OR object_closure:"MONDO:0020121")',
+            '(category:"biolink:GenotypeToDiseaseAssociation") AND (object:"MONDO:0020121" OR object_closure:"MONDO:0020121")',
         ],
         "filter_queries": [
             'subject:"MONDO:0020121" OR subject_closure:"MONDO:0020121" OR object:"MONDO:0020121" OR object_closure:"MONDO:0020121"'