Merge branch 'main' into allow-lucene-syntax

backend/pyproject.toml

@@ -20,7 +20,7 @@ python = "^3.9"
 pydantic = "^1.10.2"
 curies = "<1"
 linkml = "^1.6.1"
-prefixmaps = "^0.1.6"
+prefixmaps = "^0.1.7"
 
 requests = "^2.28.1"
 typer = "^0.7.0"

backend/src/monarch_py/datamodels/model.py

@@ -90,6 +90,10 @@ class Association(ConfiguredBaseModel):
         default_factory=list,
         description="""List of ExpandedCuries with id and url for evidence""",
     )
+    grouping_key: Optional[str] = Field(
+        None,
+        description="""A concatenation of fields used to group associations with the same essential/defining properties""",
+    )
     provided_by: Optional[str] = Field(None)
     provided_by_link: Optional[ExpandedCurie] = Field(
         None,
@@ -105,6 +109,19 @@ class Association(ConfiguredBaseModel):
     onset_qualifier: Optional[str] = Field(None)
     sex_qualifier: Optional[str] = Field(None)
     stage_qualifier: Optional[str] = Field(None)
+    qualifiers_label: Optional[str] = Field(None, description="""The name of the frequency_qualifier entity""")
+    qualifiers_namespace: Optional[str] = Field(
+        None, description="""The namespace/prefix of the frequency_qualifier entity"""
+    )
+    qualifiers_category: Optional[str] = Field(None, description="""The category of the frequency_qualifier entity""")
+    qualifiers_closure: Optional[List[str]] = Field(
+        default_factory=list,
+        description="""Field containing frequency_qualifier id and the ids of all of it's ancestors""",
+    )
+    qualifiers_closure_label: Optional[List[str]] = Field(
+        default_factory=list,
+        description="""Field containing frequency_qualifier name and the names of all of it's ancestors""",
+    )
     frequency_qualifier_label: Optional[str] = Field(None, description="""The name of the frequency_qualifier entity""")
     frequency_qualifier_namespace: Optional[str] = Field(
         None, description="""The namespace/prefix of the frequency_qualifier entity"""
@@ -250,6 +267,10 @@ class DirectionalAssociation(Association):
         default_factory=list,
         description="""List of ExpandedCuries with id and url for evidence""",
     )
+    grouping_key: Optional[str] = Field(
+        None,
+        description="""A concatenation of fields used to group associations with the same essential/defining properties""",
+    )
     provided_by: Optional[str] = Field(None)
     provided_by_link: Optional[ExpandedCurie] = Field(
         None,
@@ -265,6 +286,19 @@ class DirectionalAssociation(Association):
     onset_qualifier: Optional[str] = Field(None)
     sex_qualifier: Optional[str] = Field(None)
     stage_qualifier: Optional[str] = Field(None)
+    qualifiers_label: Optional[str] = Field(None, description="""The name of the frequency_qualifier entity""")
+    qualifiers_namespace: Optional[str] = Field(
+        None, description="""The namespace/prefix of the frequency_qualifier entity"""
+    )
+    qualifiers_category: Optional[str] = Field(None, description="""The category of the frequency_qualifier entity""")
+    qualifiers_closure: Optional[List[str]] = Field(
+        default_factory=list,
+        description="""Field containing frequency_qualifier id and the ids of all of it's ancestors""",
+    )
+    qualifiers_closure_label: Optional[List[str]] = Field(
+        default_factory=list,
+        description="""Field containing frequency_qualifier name and the names of all of it's ancestors""",
+    )
     frequency_qualifier_label: Optional[str] = Field(None, description="""The name of the frequency_qualifier entity""")
     frequency_qualifier_namespace: Optional[str] = Field(
         None, description="""The namespace/prefix of the frequency_qualifier entity"""
@@ -349,6 +383,7 @@ class Entity(ConfiguredBaseModel):
     )
     symbol: Optional[str] = Field(None)
     synonym: Optional[List[str]] = Field(default_factory=list)
+    uri: Optional[str] = Field(None, description="""The URI of the entity""")
 
 
 class FacetValue(ConfiguredBaseModel):
@@ -394,10 +429,10 @@ class Mapping(ConfiguredBaseModel):
     A minimal class to hold a SSSOM mapping
     """
 
-    subject_id: str = Field(..., description="""The first of the two entities being compared""")
+    subject_id: str = Field(...)
     subject_label: Optional[str] = Field(None, description="""The name of the subject entity""")
     predicate_id: str = Field(...)
-    object_id: Optional[str] = Field(None, description="""The second of the two entities being compared""")
+    object_id: str = Field(...)
     object_label: Optional[str] = Field(None, description="""The name of the object entity""")
     mapping_justification: Optional[str] = Field(None)
 
@@ -439,6 +474,7 @@ class Node(Entity):
     provided_by: Optional[str] = Field(None)
     symbol: Optional[str] = Field(None)
     synonym: Optional[List[str]] = Field(default_factory=list)
+    uri: Optional[str] = Field(None, description="""The URI of the entity""")
 
 
 class NodeHierarchy(ConfiguredBaseModel):
@@ -530,6 +566,7 @@ class SearchResult(Entity):
     )
     symbol: Optional[str] = Field(None)
     synonym: Optional[List[str]] = Field(default_factory=list)
+    uri: Optional[str] = Field(None, description="""The URI of the entity""")
 
 
 class SearchResults(Results):
@@ -564,10 +601,10 @@ class TermPairwiseSimilarity(PairwiseSimilarity):
     A simple pairwise similarity between two atomic concepts/terms
     """
 
-    subject_id: str = Field(..., description="""The first of the two entities being compared""")
+    subject_id: str = Field(...)
     subject_label: Optional[str] = Field(None, description="""The name of the subject entity""")
     subject_source: Optional[str] = Field(None, description="""the source for the first entity""")
-    object_id: Optional[str] = Field(None, description="""The second of the two entities being compared""")
+    object_id: str = Field(...)
     object_label: Optional[str] = Field(None, description="""The name of the object entity""")
     object_source: Optional[str] = Field(None, description="""the source for the second entity""")
     ancestor_id: Optional[str] = Field(

backend/src/monarch_py/datamodels/model.yaml

@@ -53,6 +53,7 @@ classes:
       - evidence_count
       - has_evidence
       - has_evidence_links
+      - grouping_key
       - provided_by
       - provided_by_link
       - publications
@@ -172,6 +173,7 @@ classes:
       - in_taxon_label
       - symbol
       - synonym
+      - uri
   EntityResults:
     is_a: Results
     slots:
@@ -456,6 +458,8 @@ slots:
   xref:
     multivalued: true
     range: string
+  uri:
+    description: The URI of the entity
   url:
     range: string
   subject_label:
@@ -496,6 +500,24 @@ slots:
     is_a: in_taxon
   object_taxon_label:
     is_a: in_taxon_label
+
+  # this whole block will be renamed to qualifer_ after we bring in the next biolink model release
+  qualifiers_label:
+    is_a: name
+    description: The name of the frequency_qualifier entity
+  qualifiers_namespace:
+    range: string
+    description: The namespace/prefix of the frequency_qualifier entity
+  qualifiers_category:
+    is_a: category
+    description: The category of the frequency_qualifier entity
+  qualifiers_closure:
+    multivalued: true
+    description: Field containing frequency_qualifier id and the ids of all of it's ancestors
+  qualifiers_closure_label:
+    multivalued: true
+    description: Field containing frequency_qualifier name and the names of all of it's ancestors
+
   frequency_qualifier_label:
     is_a: name
     description: The name of the frequency_qualifier entity
@@ -557,12 +579,16 @@ slots:
     multivalued: true
     description: Field containing stage_qualifier name and the names of all of it's ancestors
 # sssom slots
-  # subject_id already exists in similarity.yaml
+  subject_id:
+    range: string
+    required: true
   # subject label is already included in this schema
   predicate_id:
-      range: string
-      required: true
-  # object_id already exists in similarity.yaml
+    range: string
+    required: true
+  object_id:
+    range: string
+    required: true
   # object label is already included in this schema
   mapping_justification:
       range: string

backend/src/monarch_py/datamodels/similarity.yaml

@@ -95,22 +95,22 @@ types:
     minimum_value: 0
 
 slots:
-  subject_id:
-    slot_uri: sssom:subject_id
-    required: true
-    range: uriorcurie
-    description: The first of the two entities being compared
+#  subject_id:
+#    slot_uri: sssom:subject_id
+#    required: true
+#    range: uriorcurie
+#    description: The first of the two entities being compared
   # Excluded, since it conflicts with subject_label from this schema
   #  subject_label:
   #    slot_uri: sssom:subject_label
   #    description: the label or name for the first entity
   subject_source:
     slot_uri: sssom:subject_source
     description: the source for the first entity
-  object_id:
-    slot_uri: sssom:object_id
-    range: uriorcurie
-    description: The second of the two entities being compared
+#  object_id:
+#    slot_uri: sssom:object_id
+#    range: uriorcurie
+#    description: The second of the two entities being compared
   # Excluded, since it conflicts with object_label from this schema
   #  object_label:
   #    slot_uri: sssom:object_label

backend/src/monarch_py/implementations/solr/solr_parsers.py

@@ -21,6 +21,7 @@
 )
 from monarch_py.datamodels.solr import HistoPhenoKeys, SolrQueryResult
 from monarch_py.utils.association_type_utils import get_association_type_mapping_by_query_string
+from monarch_py.utils.entity_utils import get_uri
 from monarch_py.utils.utils import get_links_for_field, get_provided_by_link
 
 ####################
@@ -85,6 +86,7 @@ def parse_association_counts(query_result: SolrQueryResult, entity: str) -> Asso
 def parse_entity(solr_document: Dict) -> Entity:
     try:
         entity = Entity(**solr_document)
+        entity.uri = get_uri(entity.id)
     except ValidationError:
         logger.error(f"Validation error for {solr_document}")
         raise

backend/src/monarch_py/implementations/sql/sql_implementation.py

@@ -3,11 +3,13 @@
 
 import pystow
 from loguru import logger
+from pydantic import ValidationError
+
 from monarch_py.datamodels.model import Association, AssociationResults, Entity, Node, NodeHierarchy
 from monarch_py.interfaces.association_interface import AssociationInterface
 from monarch_py.interfaces.entity_interface import EntityInterface
+from monarch_py.utils.entity_utils import get_uri
 from monarch_py.utils.utils import SQL_DATA_URL, dict_factory
-from pydantic import ValidationError
 
 monarchstow = pystow.module("monarch")
 
@@ -49,6 +51,7 @@ def get_entity(self, id: str, update: bool = False, extra: bool = False) -> Enti
             "in_taxon": sql_data["in_taxon"],
             "symbol": sql_data["symbol"],
             "synonym": sql_data["synonym"].split("|"),
+            "uri": get_uri(sql_data["id"]),
         }
         try:
             results["source"] = sql_data["source"]

backend/src/monarch_py/utils/entity_utils.py

@@ -0,0 +1,6 @@
+from monarch_py.service.curie_service import CurieService
+
+
+def get_uri(id: str) -> str:
+    """Returns the URI for the given CURIE."""
+    return CurieService().expand(id)