ClinVarXMLParser: more fixes to HGVS parsing

SACGF · Sep 1, 2023 · 3c66ae5 · 3c66ae5
1 parent e5eae44
commit 3c66ae5
Show file tree

Hide file tree

Showing 4 changed files with 14 additions and 12 deletions.
diff --git a/annotation/admin.py b/annotation/admin.py
@@ -103,7 +103,7 @@ def refresh_force_vcv(self, request, queryset: QuerySet[ClinVarRecordCollection]
         messages.info(request, message=f"Fetching took {duration}")
 
 
-    @admin_action("Refresh: Force (using RCVS)")
+    @admin_action("Refresh: Force (using RCVs)")
     def refresh_force_rcvs(self, request, queryset: QuerySet[ClinVarRecordCollection]):
         start = datetime.now()
         for obj in queryset:

diff --git a/annotation/clinvar_xml_parser.py b/annotation/clinvar_xml_parser.py
@@ -69,9 +69,10 @@ def parse_xml_date(text: str) -> Optional[datetime]:
 
     @staticmethod
     def parse_hgvs(text: str) -> str:
-        if match := ClinVarXmlParser.RE_GOOD_CHGVS.match(text):
-            return match.group(1)
-        return text
+        if text:
+            if match := ClinVarXmlParser.RE_GOOD_CHGVS.match(text):
+                return match.group(1)
+        return None
 
     @classmethod
     @abstractmethod

diff --git a/annotation/clinvar_xml_parser_via_rcvs.py b/annotation/clinvar_xml_parser_via_rcvs.py
@@ -74,8 +74,8 @@ def parse_allele_origin(self, elem):
         "AttributeSet",
         PP("Attribute", Type="HGVS"))
     def parse_c_hgvs(self, elem):
-        if hgvs := elem.text:
-            self.latest.c_hgvs = ClinVarXmlParser.parse_hgvs(hgvs)
+        if hgvs := ClinVarXmlParser.parse_hgvs(elem.text):
+            self.latest.c_hgvs = hgvs
 
     @parser_path(
         PP("MeasureSet", Type="Variant"),

diff --git a/annotation/clinvar_xml_parser_via_vcv.py b/annotation/clinvar_xml_parser_via_vcv.py
@@ -7,7 +7,7 @@
 
 class ClinVarXmlParserViaVCV(ClinVarXmlParser):
 
-    PARSER_VERSION = 201  # change this whenever the parsing changes, so we know to ignore the old cache
+    PARSER_VERSION = 202  # change this whenever the parsing changes, so we know to ignore the old cache
 
     @classmethod
     def load_from_clinvar_id(cls, clinvar_variation_id: int) -> ClinVarXmlParserOutput:
@@ -79,18 +79,19 @@ def parse_clinical_significance(self, elem):
         "OtherNameList",
         "Name"
     )
-    def parse_possible_c_hgvs(self, elem):
+    def _parse_possible_c_hgvs(self, elem):
         if elem.get("Type") is None and (hgvs := elem.text):
-            self.latest.c_hgvs = ClinVarXmlParser.parse_hgvs(hgvs)
+            if hgvs := ClinVarXmlParser.parse_hgvs(hgvs):
+                self.latest.c_hgvs = hgvs
 
     @parser_path(
         "SimpleAllele",
         "AttributeSet",
         PP("Attribute", Type="HGVS"))
-    def parse_c_hgvs(self, elem):
+    def _parse_c_hgvs(self, elem):
         if not self.latest.c_hgvs:
-            if hgvs := elem.text:
-                self.latest.c_hgvs = ClinVarXmlParser.parse_hgvs(hgvs)
+            if hgvs := ClinVarXmlParser.parse_hgvs(elem.text):
+                self.latest.c_hgvs = hgvs
 
     @parser_path(
         "SimpleAllele",