Add directionality param to semsim search api (#762)

This passes the new directionality param back to semsimian-server (and
then to semsimian). No change to the UI yet.

backend/src/monarch_py/api/additional_models.py

@@ -38,6 +38,12 @@ class SemsimSearchGroup(Enum):
     MONDO = "Human Diseases"
 
 
+class SemsimDirectionality(str, Enum):
+    BIDIRECTIONAL = "bidirectional"
+    SUBJECT_TO_OBJECT = "subject_to_object"
+    OBJECT_TO_SUBJECT = "object_to_subject"
+
+
 class SemsimCompareRequest(BaseModel):
     subjects: List[str] = Field(..., title="List of subjects for comparison")
     objects: List[str] = Field(..., title="List of objects for comparison")
@@ -60,6 +66,9 @@ class SemsimSearchRequest(BaseModel):
     termset: List[str] = Field(..., title="Termset to search")
     group: SemsimSearchGroup = Field(..., title="Group of entities to search within (e.g. Human Genes)")
     metric: SemsimMetric = Field(SemsimMetric.ANCESTOR_INFORMATION_CONTENT, title="Similarity metric to use")
+    directionality: SemsimDirectionality = (
+        Query(SemsimDirectionality.BIDIRECTIONAL, title="Directionality of the search"),
+    )
     limit: Optional[int] = Field(10, title="Limit the number of results", ge=1, le=50)
 
 

backend/src/monarch_py/api/semsim.py

@@ -6,6 +6,7 @@
     SemsimSearchRequest,
     SemsimSearchGroup,
     SemsimMultiCompareRequest,
+    SemsimDirectionality,
 )
 from monarch_py.api.config import semsimian, solr
 from monarch_py.api.utils.similarity_utils import parse_similarity_prefix
@@ -124,6 +125,9 @@ def _search(
     termset: str = Path(..., title="Termset to search"),
     group: SemsimSearchGroup = Path(..., title="Group of entities to search within (e.g. Human Genes)"),
     metric: SemsimMetric = Query(SemsimMetric.ANCESTOR_INFORMATION_CONTENT, title="Similarity metric to use"),
+    directionality: SemsimDirectionality = Query(
+        SemsimDirectionality.BIDIRECTIONAL, title="Directionality of the search"
+    ),
     limit: int = Query(default=10, ge=1, le=50),
 ):
     """Search for terms in a termset
@@ -138,7 +142,9 @@ def _search(
         List[str]: List of matching terms
     """
     terms = [term.strip() for term in termset.split(",")]
-    results = semsimian().search(termset=terms, prefix=parse_similarity_prefix(group), metric=metric, limit=limit)
+    results = semsimian().search(
+        termset=terms, prefix=parse_similarity_prefix(group), metric=metric, directionality=directionality, limit=limit
+    )
     return results
 
 

backend/src/monarch_py/service/semsim_service.py

@@ -3,7 +3,7 @@
 
 from pydantic import BaseModel
 
-from monarch_py.api.additional_models import SemsimMetric, SemsimMultiCompareRequest
+from monarch_py.api.additional_models import SemsimMetric, SemsimMultiCompareRequest, SemsimDirectionality
 from monarch_py.datamodels.model import TermSetPairwiseSimilarity, SemsimSearchResult, Entity
 
 
@@ -75,10 +75,11 @@ def search(
         termset: List[str],
         prefix: str,
         metric: SemsimMetric = SemsimMetric.ANCESTOR_INFORMATION_CONTENT,
+        directionality: SemsimDirectionality = SemsimDirectionality.BIDIRECTIONAL,
         limit: int = 10,
     ) -> List[SemsimSearchResult]:
         host = f"http://{self.semsim_server_host}:{self.semsim_server_port}"
-        path = f"search/{','.join(termset)}/{prefix}/{metric}?limit={limit}"
+        path = f"search/{','.join(termset)}/{prefix}:/{metric}?limit={limit}&directionality={directionality.value} "
         url = f"{host}/{path}"
 
         print(f"Fetching {url}...")

backend/src/monarch_py/service/solr_service.py

@@ -33,7 +33,14 @@ def query(self, q: SolrQuery) -> SolrQueryResult:
         response.raise_for_status()
         solr_query_result = SolrQueryResult.model_validate(data, from_attributes=True)
         for doc in solr_query_result.response.docs:
-            self._strip_json(doc, "_version_", "iri")
+            self._strip_json(
+                doc,
+                "_version_",
+                "iri",
+                "frequency_computed_sortable_float",
+                "has_quotient_sortable_float",
+                "has_percentage_sortable_float",
+            )
 
         return solr_query_result
 

backend/tests/api/test_semsim_router.py

@@ -3,7 +3,12 @@
 from fastapi import status
 from unittest.mock import MagicMock, patch
 
-from monarch_py.api.additional_models import SemsimMetric, SemsimSearchGroup, SemsimMultiCompareRequest
+from monarch_py.api.additional_models import (
+    SemsimMetric,
+    SemsimSearchGroup,
+    SemsimMultiCompareRequest,
+    SemsimDirectionality,
+)
 from monarch_py.api.semsim import router
 from monarch_py.datamodels.category_enums import AssociationPredicate, EntityCategory
 
@@ -79,9 +84,11 @@ def test_get_search(mock_search, termset: str, metric: SemsimMetric):
     limit = 5
 
     response = client.get(f"/search/{termset}/{group.value}?metric={metric}&limit={limit}")
-
+    directionality = SemsimDirectionality.BIDIRECTIONAL
     assert response.status_code == status.HTTP_200_OK
-    mock_search.assert_called_once_with(termset=["HP:123", "HP:456"], prefix=group.name, metric=metric, limit=limit)
+    mock_search.assert_called_once_with(
+        termset=["HP:123", "HP:456"], prefix=group.name, metric=metric, directionality=directionality, limit=limit
+    )
 
 
 @patch("monarch_py.service.semsim_service.SemsimianService.search")
@@ -91,10 +98,18 @@ def test_post_search(mock_search):
     termset = ["HP:123", "HP:456"]
     group = SemsimSearchGroup.HGNC
     metric = SemsimMetric.JACCARD_SIMILARITY
+    directionality = SemsimDirectionality.BIDIRECTIONAL
     limit = 5
 
     response = client.post(
-        f"/search/", json={"termset": termset, "group": group.value, "metric": metric, "limit": limit}
+        f"/search/",
+        json={
+            "termset": termset,
+            "group": group.value,
+            "metric": metric,
+            "directionality": directionality,
+            "limit": limit,
+        },
     )
 
     assert response.status_code == status.HTTP_200_OK

backend/tests/fixtures/association_counts.py

@@ -7,10 +7,12 @@ def association_counts():
         "items": [
             {
                 "label": "Phenotype to Disease",
-                "count": 3959,
+                "count": 4012,
                 "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": 237, "category": "biolink:GenotypeToDiseaseAssociation"},
         ]
     }