CanDIG/candig-api merging: DIG-2224: Variants search (#80)

* updating submodules

* [wip] integration tests

* Fixup test_beacon_genomic

* Update integration tests

* Update candig-api to post-fixes

* Also update htsget

---------

Co-authored-by: github-actions <github-actions@github.com>
Co-authored-by: fnguyen <OrdinaryNeumann@gmail.com>

Author: 3 people

etc/tests/integration/test_integration.py

@@ -69,6 +69,33 @@ def dataset_size():
         "SITE_PM2C~SYNTH_04": 20
     }
 
+def genomic_hits_in_dataset():
+    return {
+        "SITE_PM2C~SYNTH_01": 1,
+        "SITE_PM2C~SYNTH_02": 1,
+        "SITE_PM2C~SYNTH_03": 0,
+        "SITE_PM2C~SYNTH_04": 0
+    }
+
+def genomic_variants_per_sample():
+    return {
+        "SITE_PM2C~SYNTH_02": {
+            "SAMPLE_0061": 3,
+            "SAMPLE_0062": 3
+        },
+        "SITE_PM2C~SYNTH_01": {
+            "SAMPLE_NULL_0002": 2,
+            "SAMPLE_NULL_0001": 2
+        }
+    }
+
+def genomic_and_clinical_hits_in_dataset():
+    return {
+        "SITE_PM2C~SYNTH_01": 1,
+        "SITE_PM2C~SYNTH_02": 1,
+        "SITE_PM2C~SYNTH_03": 0,
+        "SITE_PM2C~SYNTH_04": 0
+    }
 
 ## Keycloak tests:
 
@@ -683,6 +710,22 @@ def sample_request_body(filter_id, granularity="record"):
     }
 
 
+def sample_genomic_request_body(gene_id, granularity="record"):
+    return {
+        "meta": {
+            "apiVersion": "v2.0.0"
+        },
+        "query": {
+            "requestedGranularity": granularity,
+            "requestParameters": {
+                "g_variant": {
+                    "gene_id": gene_id
+                }
+            }
+        }
+    }
+
+
 @pytest.mark.parametrize("user, dataset", user_auth_datasets())
 def test_beacon_query(user, dataset):
     """Test whether a user can execute a query."""
@@ -710,15 +753,40 @@ def test_beacon_query(user, dataset):
     # Ensure that the discovery query also matches up
     assert response.json()["info"]["patients_per_program"][dataset] == dataset_size()[dataset]
 
+    # Switch to a query on variants in a given gene
+    body = sample_genomic_request_body(f"SLX9")
+    response = requests.post(
+        f"{ENV['CANDIG_URL']}/candig-api/v1/beacon/persons",
+        headers = headers,
+        json = body
+    )
+    assert response.status_code == 200
+    if user != "CANDIG_NOT_ADMIN2":
+        assert len(response.json()["response"]["resultSets"][0]["results"]) == min(10, genomic_hits_in_dataset()[dataset])
+        assert response.json()["response"]["resultSets"][0]["resultsCount"] == genomic_hits_in_dataset()[dataset]
+        assert response.json()["info"]["patients_per_program"][dataset] == genomic_hits_in_dataset()[dataset]
+
+    # Switch to a query on both gene and clinical feature
+    body = sample_request_body(f"dataset_id:{dataset}")
+    body["query"]["requestParameters"] = { "g_variant": { "gene_id": "SLX9" } }
+    response = requests.post(
+        f"{ENV['CANDIG_URL']}/candig-api/v1/beacon/persons",
+        headers = headers,
+        json = body
+    )
+    assert response.status_code == 200
+    if user != "CANDIG_NOT_ADMIN2":
+        assert len(response.json()["response"]["resultSets"][0]["results"]) == min(10, genomic_and_clinical_hits_in_dataset()[dataset])
+        assert response.json()["response"]["resultSets"][0]["resultsCount"] == genomic_and_clinical_hits_in_dataset()[dataset]
+        assert response.json()["info"]["patients_per_program"][dataset] == genomic_and_clinical_hits_in_dataset()[dataset]
+
     # Switch to a query on a specific thing
     body = sample_request_body("ICD10:C06.9")
     response = requests.post(
         f"{ENV['CANDIG_URL']}/candig-api/v1/beacon/persons",
         headers = headers,
         json = body
     )
-    print(body)
-    print(response.json())
     assert response.status_code == 200
 
     if user != "CANDIG_NOT_ADMIN2":
@@ -816,6 +884,33 @@ def test_beacon_granularity_unauth(user, dataset):
     assert response.json()["info"]["patients_per_program"][dataset] == dataset_size()[dataset]
 
 
+@pytest.mark.parametrize("user, dataset", user_auth_datasets())
+def test_beacon_genomic(user, dataset):
+    """Test whether a user can execute a genomic variants quick search."""
+    token = get_token(
+        username=ENV[f"{user}_USER"],
+        password=ENV[f"{user}_PASSWORD"],
+    )
+    headers = {
+        "Authorization": f"Bearer {token}",
+        "Content-Type": "application/json; charset=utf-8",
+    }
+    body = sample_genomic_request_body(f"LOC102723996")
+
+    response = requests.post(
+        f"{ENV['CANDIG_URL']}/candig-api/v1/beacon/g_variants",
+        headers = headers,
+        json = body
+    )
+    assert response.status_code == 200
+
+    # Ensure that the dataset is included in the result
+    # print(response.json())
+    this_dataset = response.json()["estimatedResults"][dataset]
+    for sample in this_dataset:
+        assert sample["variant_count"] == genomic_variants_per_sample()[dataset][sample["submitter_sample_id"]]
+
+
 ## Htsget tests:
 
 # def test_ingest_not_admin_htsget():