remove equivalent class from node hierarchy (always empty) (#295)

- Closes #257 - [x] fix identical repetitive subclasses - [x] fix sql Association validation error (classic relations/primary knowledge source madness) --------- Co-authored-by: Kevin Schaper <[email protected]>
monarch-initiative · Aug 29, 2023 · 3e099d5 · 3e099d5
1 parent cb45fff
commit 3e099d5
Show file tree

Hide file tree

Showing 26 changed files with 741 additions and 329 deletions.
diff --git a/backend/src/monarch_py/datamodels/model.py b/backend/src/monarch_py/datamodels/model.py
@@ -407,7 +407,6 @@ class Node(Entity):
 class NodeHierarchy(ConfiguredBaseModel):
 
     super_classes: List[Entity] = Field(default_factory=list)
-    equivalent_classes: List[Entity] = Field(default_factory=list)
     sub_classes: List[Entity] = Field(default_factory=list)
 
 

diff --git a/backend/src/monarch_py/datamodels/model.yaml b/backend/src/monarch_py/datamodels/model.yaml
@@ -210,7 +210,6 @@ classes:
   NodeHierarchy:
     slots:
       - super_classes
-      - equivalent_classes
       - sub_classes
   Results:
     abstract: true
@@ -261,12 +260,6 @@ slots:
       the object or in the object closure, the direction is backwards.
     range: AssociationDirectionEnum
     required: true
-  equivalent_classes:
-    range: Entity
-    multivalued: true
-    inlined: true
-    inlined_as_list: true
-    required: true
   evidence_count:
     description: count of supporting documents, evidence codes, and sources supplying evidence
     range: integer
@@ -371,8 +364,6 @@ slots:
     multivalued: true
   qualifiers:
     multivalued: true
-  relation:
-    range: string
   score:
     range: float
   sex_qualifier:

diff --git a/backend/src/monarch_py/implementations/solr/solr_implementation.py b/backend/src/monarch_py/implementations/solr/solr_implementation.py
@@ -96,13 +96,13 @@ def get_entity(self, id: str, extra: bool) -> Union[Node, Entity]:
 
     def _get_associated_entity(self, association: Association, this_entity: Entity) -> Entity:
         """Returns the id, name, and category of the other Entity in an Association given this_entity"""
-        if this_entity.id in association.subject_closure:
+        if this_entity.id == association.subject:
             entity = Entity(
                 id=association.object,
                 name=association.object_label,
                 category=association.object_category,
             )
-        elif this_entity.id in association.object_closure:
+        elif this_entity.id == association.object:
             entity = Entity(
                 id=association.subject,
                 name=association.subject_label,
@@ -156,13 +156,15 @@ def _get_node_hierarchy(self, entity: Entity) -> NodeHierarchy:
             NodeHierarchy: A NodeHierarchy object
         """
 
-        super_classes = self._get_associated_entities(entity, subject=entity.id, predicate="biolink:subclass_of")
-        equivalent_classes = self._get_associated_entities(entity, entity=entity.id, predicate="biolink:same_as")
-        sub_classes = self._get_associated_entities(entity, object=entity.id, predicate="biolink:subclass_of")
+        super_classes = self._get_associated_entities(
+            this_entity=entity, subject=entity.id, predicate="biolink:subclass_of"
+        )
+        sub_classes = self._get_associated_entities(
+            this_entity=entity, object=entity.id, predicate="biolink:subclass_of"
+        )
 
         return NodeHierarchy(
             super_classes=super_classes,
-            equivalent_classes=equivalent_classes,
             sub_classes=sub_classes,
         )
 

diff --git a/backend/src/monarch_py/implementations/sql/sql_implementation.py b/backend/src/monarch_py/implementations/sql/sql_implementation.py
@@ -139,13 +139,15 @@ def _get_node_hierarchy(self, entity: Entity) -> NodeHierarchy:
             NodeHierarchy: A NodeHierarchy object
         """
 
-        super_classes = self._get_associated_entities(entity, subject=entity.id, predicate="biolink:subclass_of")
-        equivalent_classes = self._get_associated_entities(entity, entity=entity.id, predicate="biolink:same_as")
-        sub_classes = self._get_associated_entities(entity, object=entity.id, predicate="biolink:subclass_of")
+        super_classes = self._get_associated_entities(
+            this_entity=entity, subject=entity.id, predicate="biolink:subclass_of"
+        )
+        sub_classes = self._get_associated_entities(
+            this_entity=entity, object=entity.id, predicate="biolink:subclass_of"
+        )
 
         return NodeHierarchy(
             super_classes=super_classes,
-            equivalent_classes=equivalent_classes,
             sub_classes=sub_classes,
         )
 
@@ -242,13 +244,12 @@ def get_associations(
                 "original_object": row["original_object"],
                 "category": row["category"],
                 "aggregator_knowledge_source": row["aggregator_knowledge_source"].split("|"),
-                "primary_knowledge_source": row["primary_knowledge_source"].split("|"),
+                "primary_knowledge_source": row["primary_knowledge_source"],
                 "publications": row["publications"].split("|"),
                 "qualifiers": row["qualifiers"].split("|"),
                 "provided_by": row["provided_by"],
-                "has_evidence": row["has_evidence"],
+                "has_evidence": row["has_evidence"].split("|"),
                 "stage_qualifier": row["stage_qualifier"],
-                "relation": row["relation"],
                 "negated": False if not row["negated"] else True,
                 "frequency_qualifier": row["frequency_qualifier"],
                 "onset_qualifier": row["onset_qualifier"],

diff --git a/backend/tests/fixtures/association_counts.py b/backend/tests/fixtures/association_counts.py
@@ -5,8 +5,20 @@
 def association_counts():
     return {
         "items": [
-            {"label": "Phenotypes", "count": 4011, "category": "biolink:DiseaseToPhenotypicFeatureAssociation"},
-            {"label": "Causal Genes", "count": 121, "category": "biolink:CausalGeneToDiseaseAssociation"},
-            {"label": "Correlated Genes", "count": 147, "category": "biolink:CorrelatedGeneToDiseaseAssociation"},
+            {
+                "label": "Phenotypes",
+                "count": 4011,
+                "category": "biolink:DiseaseToPhenotypicFeatureAssociation",
+            },
+            {
+                "label": "Causal Genes",
+                "count": 121,
+                "category": "biolink:CausalGeneToDiseaseAssociation",
+            },
+            {
+                "label": "Correlated Genes",
+                "count": 147,
+                "category": "biolink:CorrelatedGeneToDiseaseAssociation",
+            },
         ]
     }