Merge pull request #3 from monarch-initiative/monarch-phenotype-profi…

…le-ingest-mappingtable

Monarch phenotype profile ingest mappingtable

src/monarch_phenotype_profile_ingest/gene_to_phenotype_transform.py

@@ -15,13 +15,7 @@
 
 # Initiate koza app and mondo map from sssom file
 koza_app = get_koza_app("hpoa_gene_to_phenotype")
-
-try:
-    # For ingest (this data is unavailable for mock_koza)
-    mondo_map = koza_app.source.config.sssom_config.lut
-except:
-    # For testing ingest (allows mock_koza to bring in map_cache)
-    mondo_map = koza_app.get_map('mondo_map')
+mondo_map = koza_app.get_map('mondo_map')
 
 
 while (row := koza_app.get_row()) is not None:
@@ -39,11 +33,9 @@
     org_id = row["disease_id"].replace("ORPHA:", "Orphanet:")
     dis_id = org_id
     if dis_id in mondo_map:
-        if "MONDO" in mondo_map[dis_id]:
-            dis_id = mondo_map[dis_id]["MONDO"]
-
+        dis_id = mondo_map[dis_id]['subject_id']
+
     # TO DO: we may want to incorporate the original disease id somehow?
-    # TO DO: Need to add in the disease_context_qualifier information here once biolink is updated
 
     association = GeneToPhenotypicFeatureAssociation(id="uuid:" + str(uuid.uuid1()),
                                                      subject=gene_id,

src/monarch_phenotype_profile_ingest/gene_to_phenotype_transform.yaml

@@ -10,6 +10,9 @@ metadata: "src/monarch_phenotype_profile_ingest/metadata.yaml"
 files:
   - 'data/genes_to_phenotype.txt'
 
+depends_on:
+  - 'src/monarch_phenotype_profile_ingest/mondo_sssom_config.yaml'
+
 header: 0
 format: 'csv'
 delimiter: '\t'
@@ -41,13 +44,14 @@ edge_properties:
 
 transform_mode: 'flat'
 
-# For mondo disease_id mapping
+# We are using the mapping table functionality instead (may switch to this down the road...)
+# # For mondo disease_id mapping
 # sssom_config:
 #   files:
 #    - "data/mondo.sssom.tsv"
-#  subject_target_prefixes:
-#   - "MONDO"
-#  use_match:
+#   subject_target_prefixes:
+#    - "MONDO"
+#   use_match:
 #    - "exact"
 
 

src/monarch_phenotype_profile_ingest/mondo_sssom_config.yaml

@@ -0,0 +1,27 @@
+name: 'mondo_map'
+
+files:
+  - './data/mondo.sssom.tsv'
+
+format: 'csv'
+delimiter: '\t'
+header: 51
+
+filters:
+  - inclusion: 'include'
+    column: 'predicate_id'
+    filter_code: 'eq'
+    value: 'skos:exactMatch'
+
+key: object_id
+
+values:
+  - subject_id
+
+columns:
+  - subject_id
+  - subject_label
+  - predicate_id
+  - object_id
+  - object_label
+  - mapping_justification

tests/test_disease_mode_of_inheritance_transform.py

@@ -2,7 +2,7 @@
 from biolink_model.datamodel.pydanticmodel_v2 import DiseaseOrPhenotypicFeatureToGeneticInheritanceAssociation
 from koza.utils.testing_utils import mock_koza  # noqa: F401
 
-INGEST_NAME = "disease_mode_of_inheritance"
+INGEST_NAME = "hpoa_disease_mode_of_inheritance"
 INGEST_CODE = "./src/monarch_phenotype_profile_ingest/disease_mode_of_inheritance_transform.py"
 
 

tests/test_disease_to_phenotype_transform.py

@@ -2,7 +2,7 @@
 from biolink_model.datamodel.pydanticmodel_v2 import DiseaseToPhenotypicFeatureAssociation
 from koza.utils.testing_utils import mock_koza  # noqa: F401
 
-INGEST_NAME = "disease_to_phenotype"
+INGEST_NAME = "hpoa_disease_to_phenotype"
 INGEST_CODE = "./src/monarch_phenotype_profile_ingest/disease_to_phenotype_transform.py"
 
 

tests/test_gene_to_disease_transform.py

@@ -72,7 +72,7 @@ def row():
 @pytest.fixture
 def basic_g2d_entities(mock_koza, row):
     return mock_koza(
-        name="gene_to_disease",
+        name="hpoa_gene_to_disease",
         data=row,
         transform_code="./src/monarch_phenotype_profile_ingest/gene_to_disease_transform.py"
     )

tests/test_gene_to_phenotype_transform.py

@@ -12,17 +12,21 @@ def map_cache():
     Therefore, we supply the same information we would have any ways done, as a koza map instead.
     This requires a minor alteration to the ingest_transform.py script itself
     """
-    return {"mondo_map": {"MONDO:0013588":"OMIM:614129",
-                          "MONDO:0009341":"OMIM:235730",
-                          "MONDO:0013212":"OMIM:613287"}}
+    # return {"mondo_map": {"MONDO:0013588":"OMIM:614129",
+    #                       "MONDO:0009341":"OMIM:235730",
+    #                       "MONDO:0013212":"OMIM:613287"}}
+
+    return {"mondo_map": {"MONDO:0013588":{"subject_id":"OMIM:614129"}},
+                          "MONDO:0009341":{"subject_id":"OMIM:235730"},
+                          "MONDO:0013212":{"subject_id":"OMIM:613287"}}
 
 
 @pytest.fixture
 def source_name():
     """
     :return: string source name of HPOA Gene to Phenotype ingest
     """
-    return "gene_to_phenotype"
+    return "hpoa_gene_to_phenotype"
 
 
 @pytest.fixture