HPO gene to disease ingest

docs/Sources/HPOA.md

@@ -11,6 +11,34 @@ Association to "remarkable normality" may be added in the near future.
 
 The 'disease-to-mode-of-inheritance' ingest script parses 'inheritance' record information out from the annotation file.
 
+### Gene to Disease
+
+This ingest replaces the direct OMIM ingest so that we share g2d associations 1:1 with HPO. The mapping between association_type and biolink predicates shown below is the one way in which this ingest is opinionated, but attempts to be a direct translation into the biolink model.
+
+**genes_to_disease.txt** with the following fields:
+
+  - 'ncbi_gene_id'
+  - 'gene_symbol'
+  - 'association_type'
+  - 'disease_id'
+  - 'source'
+
+### Biolink Captured
+
+* biolink:DiseaseToPhenotypicFeatureAssociation
+    * id (random uuid)
+    * subject (ncbi_gene_id)
+    * predicate (association_type)
+      * MENDELIAN: `biolink:causes`
+      * POLYGENIC: `biolink:contributes_to`
+      * UNKNOWN: `biolink:gene_associated_with_condition`
+    * object (disease_id)
+    * primary_knowledge_source (source)
+      * medgen: `infores:omim`
+      * orphanet: `infores:orphanet`
+    * aggregator_knowledge_source (["infores:monarchinitiative"])
+      * also for medgen: `infores:medgen`
+
 ### Disease to Phenotype
 
 **phenotype.hpoa:** [A description of this file is found here](https://hpo-annotation-qc.readthedocs.io/en/latest/annotationFormat.html#phenotype-hpoa-format), has the following fields:

pyproject.toml

@@ -46,6 +46,7 @@ isort = "^5.9.1"
 mkdocs = "^1.3.0"
 mkdocs-material = "^8.2.9"
 pytest = "^7.1.1"
+black = "^23.3.0"
 
 [tool.poetry.scripts]
 ingest = "monarch_ingest.main:typer_app"

src/monarch_ingest/constants.py

@@ -0,0 +1,10 @@
+# knowledge sources
+INFORES_MONARCHINITIATIVE = "infores:monarchinitiative"
+INFORES_OMIM = "infores:omim"
+INFORES_ORPHANET = "infores:orphanet"
+INFORES_MEDGEN = "infores:medgen"
+
+# predicates
+BIOLINK_CAUSES = "biolink:causes"
+BIOLINK_CONTRIBUTES_TO = "biolink:contributes_to"
+BIOLINK_GENE_ASSOCIATED_WITH_CONDITION = "biolink:gene_associated_with_condition"

src/monarch_ingest/ingests.yaml

@@ -22,6 +22,8 @@ hgnc_gene:
   config: 'ingests/hgnc/gene.yaml'
 hpoa_disease_phenotype:
   config: 'ingests/hpoa/disease_phenotype.yaml'
+hpoa_gene_to_disease:
+  config: 'ingests/hpoa/gene_to_disease.yaml'
 hpoa_disease_mode_of_inheritance:
   config: 'ingests/hpoa/disease_mode_of_inheritance.yaml'
 hpoa_gene_to_phenotype:
@@ -30,8 +32,6 @@ hpoa_gene_to_phenotype:
 #   config: 'ingests/mgi/publication_to_gene.yaml'
 ncbi_gene:
   config: 'ingests/ncbi/gene.yaml'
-omim_gene_to_disease:
-  config: 'ingests/omim/gene_to_disease.yaml'
 panther_genome_orthologs:
   config: 'ingests/panther/genome_orthologs.yaml'
 pombase_gene:

src/monarch_ingest/ingests/hpoa/gene_to_disease.py

@@ -0,0 +1,26 @@
+from biolink.pydanticmodel import GeneToDiseaseAssociation
+from koza.cli_runner import get_koza_app
+
+from monarch_ingest.constants import INFORES_MONARCHINITIATIVE
+from monarch_ingest.ingests.hpoa.hpoa_utils import get_knowledge_sources, get_predicate
+
+koza_app = get_koza_app("hpoa_gene_to_disease")
+
+
+while (row := koza_app.get_row()) is not None:
+    gene_id = row["ncbi_gene_id"]
+    disease_id = row["disease_id"]
+
+    predicate = get_predicate(row["association_type"])
+
+    primary_knowledge_source, aggregator_knowledge_source = get_knowledge_sources(row["source"], INFORES_MONARCHINITIATIVE)
+
+    association = GeneToDiseaseAssociation(
+        subject=gene_id,
+        predicate=predicate,
+        object=disease_id,
+        primary_knowledge_source=primary_knowledge_source,
+        aggregator_knowledge_source=aggregator_knowledge_source
+    )
+
+    koza_app.write(association)

src/monarch_ingest/ingests/hpoa/gene_to_disease.yaml

@@ -0,0 +1,27 @@
+name: 'hpoa_gene_to_disease'
+
+files:
+#  - './data/hpoa/genes_to_disease.txt'
+  - './data/hpoa/subset_genes_to_disease.txt'
+
+format: 'csv'
+
+delimiter: '\t'
+
+columns:
+  - 'ncbi_gene_id'
+  - 'gene_symbol'
+  - 'association_type'
+  - 'disease_id'
+  - 'source'
+
+edge_properties:
+  - 'id'
+  - 'category'
+  - 'subject'
+  - 'predicate'
+  - 'object'
+  - 'aggregator_knowledge_source'
+  - 'primary_knowledge_source'
+
+transform_mode: 'flat'

src/monarch_ingest/ingests/hpoa/hpoa_utils.py

@@ -6,6 +6,9 @@
 
 from loguru import logger
 
+from monarch_ingest.constants import INFORES_MEDGEN, INFORES_OMIM, INFORES_ORPHANET, BIOLINK_CAUSES, \
+    BIOLINK_CONTRIBUTES_TO, BIOLINK_GENE_ASSOCIATED_WITH_CONDITION
+
 
 class FrequencyHpoTerm(NamedTuple):
     curie: str
@@ -94,3 +97,39 @@ def phenotype_frequency_to_hpo_term(
         return None
 
     return hpo_term, percentage, quotient   # percentage and/or quotient will also be None if not applicable
+
+
+def get_knowledge_sources(original_source: str, additional_source: str) -> (str, List[str]):
+    """
+    Return a tuple of the primary_knowledge_source and original_knowledge_source
+    """
+    _primary_knowledge_source: str = ""
+    _aggregator_knowledge_source: List[str] = []
+
+    if additional_source is not None:
+        _aggregator_knowledge_source.append(additional_source)
+
+    if "medgen" in original_source:
+        _aggregator_knowledge_source.append(INFORES_MEDGEN)
+        _primary_knowledge_source = INFORES_OMIM
+    elif "orphadata" in original_source:
+        _primary_knowledge_source = INFORES_ORPHANET
+
+    if _primary_knowledge_source == "":
+        raise ValueError(f"Unknown knowledge source: {original_source}")
+
+    return _primary_knowledge_source, _aggregator_knowledge_source
+
+
+def get_predicate(original_predicate: str) -> str:
+    """
+    Convert the association column into a Biolink Model predicate
+    """
+    if original_predicate == 'MENDELIAN':
+        return BIOLINK_CAUSES
+    elif original_predicate == 'POLYGENIC':
+        return BIOLINK_CONTRIBUTES_TO
+    elif original_predicate == 'UNKNOWN':
+        return BIOLINK_GENE_ASSOCIATED_WITH_CONDITION
+    else:
+        raise ValueError(f"Unknown predicate: {original_predicate}")