Merge pull request #78 from Clinical-Genomics/clinsig_ignorecase

Clinsig fixcase

CHANGELOG.md

@@ -1,3 +1,7 @@
+## [unreleased]
+### Changed
+- Fix uppercase/lowercase letters when parsing clinsig terms from files
+
 ## [2.2]
 ### Changed
 - Provide the URL to the prod service running on the Clinical Genomics server instead of the stage one on README page

preClinVar/constants.py

@@ -1,6 +1,26 @@
 DRY_RUN_SUBMISSION_URL = "https://submit.ncbi.nlm.nih.gov/api/v1/submissions/?dry-run=true"
 VALIDATE_SUBMISSION_URL = "https://submit.ncbi.nlm.nih.gov/apitest/v1/submissions"
 
+CLNSIG_TERMS = [
+    "Pathogenic",
+    "Likely pathogenic",
+    "Uncertain significance",
+    "Likely benign",
+    "Benign",
+    "Pathogenic, low penetrance",
+    "Uncertain risk allele",
+    "Likely pathogenic, low penetrance",
+    "Established risk allele",
+    "Likely risk allele",
+    "affects",
+    "association",
+    "drug response",
+    "confers sensitivity",
+    "protective",
+    "other",
+    "not provided",
+]
+
 CONDITIONS_MAP = {
     "HPO": "HP",
     "MedGen": "MedGen",

preClinVar/file_parser.py

@@ -3,7 +3,7 @@
 from csv import DictReader
 from tempfile import NamedTemporaryFile
 
-from preClinVar.constants import CONDITIONS_MAP, SNV_COORDS, SV_COORDS
+from preClinVar.constants import CLNSIG_TERMS, CONDITIONS_MAP, SNV_COORDS, SV_COORDS
 
 LOG = logging.getLogger("uvicorn.access")
 
@@ -45,6 +45,12 @@ def set_item_clin_sig(item, variant_dict):
     """
     # set first required params
     clinsig = variant_dict.get("Clinical significance")
+    # Make sure clinsig term is compliant with API standards:
+    for term in CLNSIG_TERMS:
+        if clinsig.lower() == term.lower():
+            clinsig = term
+            break
+
     clinsig_comment = variant_dict.get("Comment on clinical significance")
     last_eval = variant_dict.get("Date last evaluated")
     inherit_mode = variant_dict.get("Mode of inheritance")

tests/test_file_parser.py

@@ -0,0 +1,15 @@
+from preClinVar.constants import CLNSIG_TERMS
+from preClinVar.file_parser import set_item_clin_sig
+
+
+def test_set_item_clin_sig_fix_case():
+    """Test the function that collects clinsig from Variant file when clisnsig term has wrong uppercase/lowercase"""
+    # GIVEN a variant dictionary with a non-compliant slinsig value
+    item = {}
+    variant_dict = {"Clinical significance": "Likely Pathogenic"}
+
+    # WHEN clisig is collected from variant_dict
+    set_item_clin_sig(item, variant_dict)
+
+    # THEN it's converted into a compliant term
+    assert item["clinicalSignificance"]["clinicalSignificanceDescription"] in CLNSIG_TERMS