add optional separate genomic_id for variantfiles

data/files.sql

@@ -149,6 +149,7 @@ INSERT INTO alias VALUES('y','Y');
 INSERT INTO alias VALUES('mt','MT');
 CREATE TABLE variantfile (
 	id VARCHAR NOT NULL, 
+	genomic_id VARCHAR,
 	drs_object_id VARCHAR, 
 	indexed INTEGER,
 	chr_prefix VARCHAR,

htsget_server/database.py

@@ -91,6 +91,7 @@ class Contig(ObjectDBBase):
 class VariantFile(ObjectDBBase):
     __tablename__ = 'variantfile'
     id = Column(String, primary_key=True)
+    genomic_id = Column(String)
     indexed = Column(Integer)
     chr_prefix = Column(String)
     reference_genome = Column(String)
@@ -130,6 +131,7 @@ def __repr__(self):
         result = {
             'id': self.id,
             'drsobject': self.drs_object_id,
+            'genomic_id': self.genomic_id,
             'indexed': self.indexed,
             'chr_prefix': self.chr_prefix,
             'reference_genome': self.reference_genome,
@@ -497,14 +499,18 @@ def get_variantfile(variantfile_id):
 
 
 def create_variantfile(obj):
-    # obj = {"id", "reference_genome"}
+    # obj = {"id", "reference_genome", "genomic_id"}
     with Session() as session:
         new_variantfile = session.query(VariantFile).filter_by(id=obj['id']).one_or_none()
         if new_variantfile is None:
             new_variantfile = VariantFile()
             new_variantfile.indexed = 0
             new_variantfile.chr_prefix = '0'
         new_variantfile.id = obj['id']
+        if "genomic_id" in obj:
+            new_variantfile.genomic_id = obj['genomic_id']
+        else:
+            new_variantfile.genomic_id = obj['id']
         new_variantfile.reference_genome = obj['reference_genome']
         new_drs = session.query(DrsObject).filter_by(id=obj['id']).one_or_none()
         if new_drs is not None:

htsget_server/openapi.yml

@@ -191,6 +191,7 @@ paths:
                 - $ref: '#/components/parameters/idPathParam'
                 - $ref: '#/components/parameters/forceParam'
                 - $ref: '#/components/parameters/refGenomeParam'
+                - $ref: '#/components/parameters/genomicIdParam'
             responses:
                 200:
                     description: Successfully indexed variant file
@@ -666,6 +667,13 @@ components:
             required: false
             schema:
                 $ref: '#/components/schemas/ReferenceGenome'
+        genomicIdParam:
+            in: query
+            name: genomic_id
+            description: If specified, sets this variant entity's associated genomic_id. Otherwise default is the same as the id.
+            required: false
+            schema:
+                type: string
         readsFormatParam:
             in: query
             name: format

htsget_server/operations.py

@@ -111,11 +111,14 @@ def get_variants_data(id_, reference_name=None, format_="VCF", start=None, end=N
 
 
 @app.route('/variants/<path:id_>/index')
-def index_variants(id_=None, force=False, genome='hg38'):
+def index_variants(id_=None, force=False, genome='hg38', genomic_id=None):
     if not authz.is_site_admin(request):
         return {"message": "User is not authorized to index variants"}, 403
     if id_ is not None:
-        varfile = database.create_variantfile({"id": id_, "reference_genome": genome})
+        params = {"id": id_, "reference_genome": genome}
+        if genomic_id is not None:
+            params["genomic_id"] = genomic_id
+        varfile = database.create_variantfile(params)
         if varfile is not None:
             if varfile['indexed'] == 1 and not force:
                 return varfile, 200
@@ -193,6 +196,7 @@ def search_variants():
             htsget_obj['urls'].append(_get_base_url("variant", drs_obj_id, data=False))
             htsget_obj['id'] = drs_obj_id
             htsget_obj['variantcount'] = count
+            htsget_obj['genomic_id'] = database.get_variantfile(drs_obj_id)['genomic_id']
             htsget_obj['samples'] = database.get_samples_in_drs_objects({'drs_object_ids': [drs_obj_id]})
             htsget_obj['reference_genome'] = searchresult['reference_genome'][i]
             result['results'].append(htsget_obj)

tests/test_htsget_server.py

@@ -78,17 +78,21 @@ def test_post_update():
 
 
 def test_index_variants():
-    return [('sample.compressed'), ('NA18537'), ('NA20787'), ('multisample_1'), ('multisample_2')]
+    return [('sample.compressed', None), ('NA18537', None), ('NA20787', None), ('multisample_1', 'HG00096'), ('multisample_2', 'HG00097')]
 
 
-@pytest.mark.parametrize('sample', test_index_variants())
-def test_index_variantfile(sample):
+@pytest.mark.parametrize('sample, genomic_id', test_index_variants())
+def test_index_variantfile(sample, genomic_id):
     url = f"{HOST}/htsget/v1/variants/{sample}/index"
     params = {"genome": "hg37"}
+    if genomic_id is not None:
+      params["genomic_id"] = genomic_id
     #params['force'] = True
     response = requests.get(url, params=params, headers=headers)
     print(response.text)
     assert response.json()["id"] == sample
+    if genomic_id is not None:
+      assert response.json()["genomic_id"] == genomic_id
 
 
 def invalid_start_end_data():