Fetch genotypes from virtuoso

* gn3/api/metadata.py: Import get_genotype_metadata.
(genotype): New end-point.
* gn3/db/rdf.py (get_phenotype_metadata): New function.

Signed-off-by: Munyoki Kilyungi <me@bonfacemunyoki.com>

2 files changed, 59 insertions, 0 deletions

diff --git a/gn3/api/metadata.py b/gn3/api/metadata.py
index a2eff05..f21739a 100644
--- a/gn3/api/metadata.py
+++ b/gn3/api/metadata.py
@@ -11,6 +11,7 @@ from SPARQLWrapper import SPARQLWrapper
 from gn3.db.rdf import get_dataset_metadata
 from gn3.db.rdf import get_publication_metadata
 from gn3.db.rdf import get_phenotype_metadata
+from gn3.db.rdf import get_genotype_metadata
 from gn3.db.rdf import sparql_query
 from gn3.db.rdf import RDF_PREFIXES
 
@@ -67,6 +68,21 @@ def phenotype(name):
         return jsonify({})
 
 
+@metadata.route("/genotype/<name>", methods=["GET"])
+def genotype(name):
+    """Fetch a genotype's metadata given it's name"""
+    try:
+        return jsonify(
+            get_genotype_metadata(
+                SPARQLWrapper(current_app.config.get("SPARQL_ENDPOINT")),
+                name,
+            ).data
+        )
+    # The virtuoso server is misconfigured or it isn't running at all
+    except (RemoteDisconnected, URLError):
+        return jsonify({})
+
+
 @metadata.route("/genewiki/<symbol>", methods=["GET"])
 def get_genewiki_entries(symbol):
     """Fetch the GN and NCBI GeneRIF entries"""
diff --git a/gn3/db/rdf.py b/gn3/db/rdf.py
index 83bc46b..ad2e3be 100644
--- a/gn3/db/rdf.py
+++ b/gn3/db/rdf.py
@@ -26,6 +26,7 @@ PREFIX up: <http://purl.uniprot.org/core/>
 PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
 PREFIX publication: <http://genenetwork.org/publication/>
 PREFIX phenotype: <http://genenetwork.org/phenotype/>
+PREFIX genotype: <http://genenetwork.org/genotype/>
 """
 
 
@@ -243,3 +244,45 @@ CONSTRUCT {
     )[0].items():
         result[key] = value
     return result
+
+
+def get_genotype_metadata(
+        sparql_conn: SPARQLWrapper, name: str
+):
+    """Return info about a phenotype with a given NAME"""
+    __metadata_query = """
+$prefix
+
+CONSTRUCT {
+    ?genotype ?pPredicate ?pValue .
+    ?genotype gn:speciesName ?speciesName .
+    ?genotype gn:inbredSetName ?inbredSetBinomialName .
+    ?genotype gn:datasetName ?datasetFullName .
+} WHERE {
+    ?genotype ?pPredicate ?pValue .
+    OPTIONAL {
+        ?genotype gn:genotypeOfDataset ?dataset .
+        ?dataset gn:fullName ?datasetFullName .
+    }.
+    OPTIONAL {
+        ?genotype gn:genotypeOfDataset ?dataset .
+        ?dataset gn:datasetOfInbredSet ?inbredSet .
+        ?inbredSet gn:binomialName ?inbredSetBinomialName .
+        ?inbredSet gn:inbredSetOfSpecies ?species .
+        ?species gn:displayName ?speciesName .
+    } .
+    FILTER( ?genotype = genotype:$name ) .
+    MINUS {
+         ?genotype rdf:type ?pValue .
+    }
+}
+"""
+    result: MonadicDict = MonadicDict()
+    for key, value in sparql_query(
+            sparql_conn,
+            Template(__metadata_query)
+            .substitute(name=name,
+                        prefix=RDF_PREFIXES)
+    )[0].items():
+        result[key] = value
+    return result