Add HLA normalization via py-ard

Author: mihaitodor

app/api_spec.yml

@@ -1225,6 +1225,28 @@ paths:
             type: string
             example: "NM_021960.4:c.740C>T"
 
+  /utilities/normalize-hla:
+    get:
+      summary: "Normalize HLA"
+      operationId: "app.utilities_endpoints.normalize_hla"
+      tags:
+        - "Operations Utilities (not part of balloted HL7 Operations)"
+      responses:
+        "200":
+          description: "Returns a normalized HLA ARD allele."
+          content:
+            application/json:
+              schema:
+                type: object
+      parameters:
+        - name: allele
+          in: query
+          required: true
+          description: "Allele."
+          schema:
+            type: string
+            example: "B14"
+
 tags:
   - name: Subject Genotype Operations
   - name: Subject Phenotype Operations

app/utilities_endpoints.py

@@ -1,8 +1,14 @@
+import os
+
 from flask import abort, jsonify
 from collections import OrderedDict
 from app import common
 from app import input_normalization
 from utilities import SPDI_Normalization
+import pyard
+
+pyard_database_version = os.getenv('PYARD_DATABASE_VERSION', '3550')
+ard = pyard.init(data_dir='./pyard', imgt_version=pyard_database_version)
 
 
 def get_feature_coordinates(
@@ -208,3 +214,22 @@ def normalize_variant(variant):
     except Exception as err:
         print(f"Unexpected {err=}, {type(err)=}")
         abort(422, 'Failed LiftOver')
+
+
+def normalize_hla(allele):
+    try:
+        return {
+            allele: {
+                "G": ard.redux(allele, "G"),
+                "P": ard.redux(allele, "P"),
+                "lg": ard.redux(allele, "lg"),
+                "lgx": ard.redux(allele, "lgx"),
+                "W": ard.redux(allele, "W"),
+                "exon": ard.redux(allele, "exon"),
+                "U2": ard.redux(allele, "U2"),
+                "S": ard.redux(allele, "S")
+            }
+        }
+    except Exception as err:
+        print(f"Unexpected {err=}, {type(err)=}")
+        abort(422, 'Failed HLA normalization')

fetch_refseq.sh

@@ -21,3 +21,13 @@ tar -xzf GRCh38_refseq.tar.gz
 tar -xzf rna_refseq.tar.gz
 
 echo "Finished extracting refseq files."
+
+# TODO: REFACTOR THIS!
+cd ..
+mkdir -p ./pyard
+cd ./pyard
+curl -sLO https://github.com/FHIR/genomics-operations/releases/download/113c119/pyard.sqlite3.tar.gz
+
+echo "Extracting pyard database..."
+tar -xzf pyard.sqlite3.tar.gz
+echo "Finished extracting pyard database."

requirements.txt

@@ -6,7 +6,8 @@ Flask==2.2.2
 flask_cors==4.0.0
 gunicorn==20.1.0
 hgvs==1.5.4
-pandas==1.3.5
+pandas==1.5.3
+py-ard==1.0.10
 pyfastx==2.0.1
 pyliftover==0.4
 pymongo==4.2.0