Merge pull request #1549 from BRCAChallenge/enigma_cleanup

Enigma cleanup

Author: melissacline

pipeline/clinvar/clinvar_common.py

@@ -17,6 +17,7 @@ def isCurrent(element):
     else:
         return(rr.text == "current")
 
+    
 def textIfPresent(element, field):
     """Return the text associated with a field under the element, or
     None if the field is not present"""
@@ -28,6 +29,7 @@ def textIfPresent(element, field):
     else:
         return None
 
+    
 def findUniqueElement(name, parent):
     """Find a child element directly or indirectly underneath this parent
        element which should occur only once (i.e. there should be no other
@@ -256,13 +258,14 @@ def __init__(self, element, debug=False):
         self.element = element        
         self.condition_type = None
         self.condition_value = None
+        self.condition_category = None
         self.condition_db_id = list()
         for trait in element.iter("Trait"):
             self.condition_type = trait.get("Type")
             for name in trait.iter("Name"):
                 ev = name.find("ElementValue")
                 if ev.get("Type") == "Preferred":
-                    self.condition_value = ev.text
+                    self.condition_category = ev.text
                 for xref in trait.iter("XRef"):
                     if not re.search("Genetic Testing Registry", xref.get("DB")):
                         xref_string = xref.get("DB") + "_" + xref.get("ID")

pipeline/clinvar/enigma_from_clinvar.py

@@ -18,20 +18,14 @@
 MULTI_ENTRY_SEP = ','
 
 
-
 three_letters_aa = re.compile('p.\\(?[A-Z][a-z]{2}[0-9]+[A-Z][a-z]{2}') # e.g. p.(Tyr831SerfsTer9)
+
 def _is_bic_designation(s):
     return any(k in s.lower() for k in {'ins', 'del', 'dup'}) or \
         (not s.startswith('p.') and '>' in s) or \
         (s.startswith('p.') and ':' not in s and three_letters_aa.match(s) is None) # shouldn't match for a BIC designator
 
 
-#
-# 11/27/2024: ClinVar no longer supplies the BIC nomenclature terms in their records
-def _fetch_bic(cvs_el):
-    return default_val
-
-
 def _compute_protein_changes(hgvs_cdna, hgvs_util):
     if hgvs_cdna is not None:
         v_protein = hgvs_util.cdna_to_protein(hgvs_cdna, return_str=False) 
@@ -53,6 +47,17 @@ def _compute_protein_changes(hgvs_cdna, hgvs_util):
     return (default_val, default_val)
 
 
+def _fetch_bic(va_el):
+    bic_list = list()
+    for name_item in va_el.findall("./ClassifiedRecord/SimpleAllele/OtherNameList/Name"):
+        name = name_item.text
+        if _is_bic_designation(name):
+            bic_list.append(name)
+    if len(bic_list) > 0:
+        return '|'.join(bic_list)
+    return default_val
+
+
 
 
 
@@ -79,8 +84,8 @@ def _parse_engima_assertion(enigma_assertion, hgvs_util):
     rec["Assertion_method"] = enigma_assertion.assertionMethod
     rec["Assertion_method_citation"] = enigma_assertion.assertionMethodCitation
     rec["Comment_on_clinical_significance"] = enigma_assertion.summaryEvidence
-    rec["Collection_method"] = ','.join(enigma_assertion.method.capitalize())
-    rec["Allele_origin"] = ','.join(enigma_assertion.origin.capitalize())
+    rec["Collection_method"] = ','.join(enigma_assertion.method).capitalize()
+    rec["Allele_origin"] = ','.join(enigma_assertion.origin).capitalize()
     rec["ClinVarAccession"] = "%s.%s" % (enigma_assertion.accession,
                                          enigma_assertion.accession_version)
     return rec
@@ -155,13 +160,17 @@ def parse_record(va_el, hgvs_util, symbols, mane_transcript,
             rec["Reference_sequence"] = transcript.split(":")[0]
             rec["HGVS_cDNA"] = transcript.split(":")[1]
         rec["Condition_ID_type"] = va.classification.condition_type
-        if va.classification_condition_value == None:
+        if va.classification.condition_value == None:
             rec["Condition_ID_value"] = "not provided"
         else:
             rec["Condition_ID_value"] = va.classification.condition_value
         #
-        # 11/25/2024: the trait set / condition category has been deprecated (is no longer displayed)
-        rec["Condition_category"] = default_val
+        # 2/28/25: it seems counter-intuitive that the condition category
+        # comes from the ClinVar condition type field, while the condition
+        # type is blank, but that's the way it is.  FWIW, neither of these
+        # fields are actually displayed.
+        rec["Condition_category"] = va.classification.condition_type
+        rec["Condition_type"] = default_val
         for scv_accession in va.otherAssertions.keys():
             oa = va.otherAssertions[scv_accession]
             if oa.reviewStatus == "reviewed by expert panel":