Merge pull request #23 from bcgsc/bugfix/KBDEV-1231_wildtype_mismatch

Bugfix/kbdev 1231 wildtype mismatch

pori_python/ipr/main.py

@@ -237,8 +237,8 @@ def ipr_report(
     always_write_output_json: bool = False,
     ipr_upload: bool = True,
     interactive: bool = False,
-    graphkb_username: str = None,
-    graphkb_password: str = None,
+    graphkb_username: str = "",
+    graphkb_password: str = "",
     graphkb_url: str = "",
     generate_therapeutics: bool = False,
     generate_comments: bool = True,

tests/test_graphkb/test_match.py

@@ -485,6 +485,12 @@ def test_genomic_coordinates(self, conn):
     def test_tert_promoter(self, conn):
         assert match.match_positional_variant(conn, "TERT:c.-124C>T")
 
+    def test_wildtype_match_error(self, conn):
+        for gkb_match in match.match_positional_variant(conn, "TP53:p.E285K"):
+            assert (
+                "wildtype" not in gkb_match["displayName"]
+            ), f"TP53:p.E285K should not match {gkb_match['displayName']}"
+
     @pytest.mark.skipif(
         True, reason="GERO-303 - technically incorrect notation for GSC backwards compatibility."
     )

tests/test_ipr/test_annotate.py

@@ -41,6 +41,39 @@
     ),
 }
 
+KBDEV1231_TP53_ERR_MATCH_WT = {
+    "altSeq": "",
+    "chromosome": "chr17",
+    "comments": "",
+    "endPosition": "",
+    "gene": "TP53",
+    "germline": False,
+    "hgvsCds": "ENST00000269305:c.853G>A",
+    "hgvsGenomic": "chr17:g.7673767C>T",
+    "hgvsProtein": "TP53:p.E285K",
+    "key": "c23a7b0387335e7a5ed6c1081a1822ae",
+    "library": "F145233;F145265",
+    "ncbiBuild": "GRCh38",
+    "normalAltCount": "",
+    "normalDepth": "",
+    "normalRefCount": "",
+    "proteinChange": "p.E285K",
+    "refSeq": "",
+    "rnaAltCount": 311,
+    "rnaDepth": 370,
+    "rnaRefCount": 59,
+    "startPosition": "",
+    "transcript": "ENST00000269305",
+    "tumourAltCopies": "",
+    "tumourAltCount": 64,
+    "tumourDepth": 100,
+    "tumourRefCopies": "",
+    "tumourRefCount": 36,
+    "variant": "TP53:p.E285K",
+    "variantType": "mut",
+    "zygosity": "",
+}
+
 
 @pytest.fixture(scope="module")
 def graphkb_conn():
@@ -106,3 +139,11 @@ def test_annotate_structural_variants_tp53(self, graphkb_conn):
             missing = pref_vars.difference(alt_vars).difference(known_issues)
             print(alt_vars)
             assert not missing, f"{key} missing{missing}: {diff}"
+
+    def test_wt_not_matched(self, graphkb_conn):
+        """Verify wildtypes are not matched to mutations."""
+        disease = "cancer"
+        matches = annotate_positional_variants(graphkb_conn, [KBDEV1231_TP53_ERR_MATCH_WT], disease)
+        # KBDEV-1231 - wildtype - should not match.  A mutation is not wildtype
+        wt_matches = sorted(set([m["kbVariant"] for m in matches if "wildtype" in m["kbVariant"]]))
+        assert not wt_matches, f"Mutation 'TP53:p.E285K' should NOT match {wt_matches}"

tests/test_ipr/test_upload.py

@@ -4,8 +4,8 @@
 import pytest
 import sys
 import uuid
-from typing import Dict, Tuple, List
-from unittest.mock import MagicMock, patch
+from typing import Generator
+from unittest.mock import patch
 
 from pori_python.ipr.connection import IprConnection
 from pori_python.ipr.main import command_interface
@@ -31,7 +31,7 @@ def get_test_file(name: str) -> str:
 
 
 @pytest.fixture(scope="module")
-def loaded_reports(tmp_path_factory) -> Dict:
+def loaded_reports(tmp_path_factory) -> Generator:
     json_file = tmp_path_factory.mktemp("inputs") / "content.json"
     async_json_file = tmp_path_factory.mktemp("inputs") / "async_content.json"
     patient_id = f"TEST_{str(uuid.uuid4())}"