issue #1137 - clingen swap reference base

genes/hgvs/hgvs_converter.py

@@ -17,10 +17,17 @@ def is_internal_type(self) -> bool:
 
 
 class HgvsMatchRefAllele:
-    def __init__(self, provided_ref: str, calculated_ref: str):
+    def __init__(self, provided_ref: str, calculated_ref: str, ref_type=None, ref_source=None):
         self.provided_ref = provided_ref
         self.calculated_ref = calculated_ref
 
+        if ref_type is None:
+            ref_type = "genomic"
+        self.ref_type = ref_type
+        if ref_source is None:
+            ref_source = "our build"
+        self.ref_source = ref_source
+
     def __bool__(self):
         if self.provided_ref:
             return self.provided_ref == self.calculated_ref
@@ -40,6 +47,8 @@ def __eq__(self, other):
             return True
         return self.provided_ref == other.provided_ref and self.calculated_ref == other.calculated_ref
 
+    def get_message(self) -> str:
+        return f'Using {self.ref_type} reference "{self.calculated_ref}" from {self.ref_source}, in place of provided reference "{self.provided_ref}"'
 
 # We need a common Exception
 # Common HGVS Extra??

genes/hgvs/hgvs_matcher.py

@@ -145,7 +145,7 @@ def __init__(self, genome_build: GenomeBuild, hgvs_converter_type=None,
                                                            local_resolution=local_resolution,
                                                            clingen_resolution=clingen_resolution)
 
-    def _clingen_get_variant_coordinate(self, hgvs_string: str) -> VariantCoordinate:
+    def _clingen_get_variant_coordinate_and_matches_reference(self, hgvs_string: str, match_ref_allele=None) -> tuple[VariantCoordinate, bool]:
         cleaned_hgvs = self.hgvs_converter.c_hgvs_remove_gene_symbol(hgvs_string)
 
         try:
@@ -158,7 +158,9 @@ def _clingen_get_variant_coordinate(self, hgvs_string: str) -> VariantCoordinate
                     start=variant_coord.start,
                     ref=variant_coord.ref,
                     alt=variant_coord.ref)  # ref == alt
-            return variant_coord
+            if match_ref_allele is None:
+                match_ref_allele = True
+            return variant_coord, match_ref_allele
         except ClinGenAlleleAPIException:
             self.attempt_clingen = False
             raise
@@ -168,6 +170,23 @@ def _clingen_get_variant_coordinate(self, hgvs_string: str) -> VariantCoordinate
                 transcript_accession = self.hgvs_converter.get_transcript_accession(hgvs_string)
                 self._set_clingen_allele_registry_missing_transcript(transcript_accession)
             else:
+                if settings.CLINGEN_ALLELE_REGISTRY_REATTEMPT_WITH_ACTUAL_REF and match_ref_allele is None:
+                    # Don't do if already swapped (stop infinite recursion)
+                    rjson = cga_se.response_json
+                    if rjson["errorType"] == 'IncorrectReferenceAllele':
+                        actual_allele = rjson['actualAllele']
+                        given_allele = rjson['givenAllele']
+                        transcript_reference_sequence = rjson["referenceSequence"]
+                        hgvs_variant = self.create_hgvs_variant(cleaned_hgvs)
+                        if hgvs_variant.ref_allele == given_allele:
+                            hgvs_variant.ref_allele = actual_allele
+                            hgvs_swapped_ref = hgvs_variant.format()
+                            match_ref_allele = HgvsMatchRefAllele(provided_ref=given_allele,
+                                                                  calculated_ref=actual_allele,
+                                                                  ref_type=f"transcript {transcript_reference_sequence}",
+                                                                  ref_source="ClinGen Allele Registry")
+                            return self._clingen_get_variant_coordinate_and_matches_reference(hgvs_swapped_ref, match_ref_allele=match_ref_allele)
+
                 self.attempt_clingen = False
             raise
 
@@ -194,10 +213,9 @@ def _lrg_get_variant_coordinate_used_transcript_method_and_matches_reference(sel
 
         hgvs_string = hgvs_variant.format()
         try:
-            # ClinGen fails if reference base is different so matches_reference is always True
-            matches_reference = True
             method = HGVSConverterType.CLINGEN_ALLELE_REGISTRY.name
-            return self._clingen_get_variant_coordinate(hgvs_string), lrg_transcript_accession, HGVSConverterType.CLINGEN_ALLELE_REGISTRY, method, matches_reference
+            variant_coordinate, matches_reference = self._clingen_get_variant_coordinate_and_matches_reference(hgvs_string)
+            return variant_coordinate, lrg_transcript_accession, HGVSConverterType.CLINGEN_ALLELE_REGISTRY, method, matches_reference
         except ClinGenAllele.ClinGenAlleleRegistryException as cga_re:
             raise ValueError(f"Could not retrieve {hgvs_string} from ClinGen Allele Registry") from cga_re
 
@@ -356,8 +374,7 @@ def get_variant_coordinate_used_transcript_kind_method_and_matches_reference(sel
                     if self._clingen_allele_registry_ok(tv.accession):
                         error_message = f"Could not convert \"{hgvs_string}\" using ClinGenAllele Registry"
                         try:
-                            matches_reference = True  # ClnGen fails if different
-                            variant_coordinate = self._clingen_get_variant_coordinate(hgvs_string_for_version)
+                            variant_coordinate, matches_reference = self._clingen_get_variant_coordinate_and_matches_reference(hgvs_string_for_version)
                         except ClinGenAlleleServerException as cga_se:
                             # If it's unknown reference we can just retry with another version, other errors are fatal
                             if cga_se.is_unknown_reference():

snpdb/signals/variant_search.py

@@ -537,7 +537,8 @@ def _search_hgvs(hgvs_string: str, user: User, genome_build: GenomeBuild, visibl
             # reporting on the "provided" reference is slightly promblematic as it's not always provided directly, it could be indirectly
 
             if isinstance(matches_reference, HgvsMatchRefAllele) and matches_reference.provided_ref:
-                search_messages.append(SearchMessage(f'Using genomic reference "{matches_reference.calculated_ref}" from our build, in place of provided reference "{matches_reference.provided_ref}"', LogLevel.ERROR, substituted=True))
+                msg = matches_reference.get_message()
+                search_messages.append(SearchMessage(msg, LogLevel.ERROR, substituted=True))
             else:
                 # if no reference was provided, do we even need to provide a message?
                 # e.g. this is providing a ref for when we have a delins, e.g. delinsGT => delCCinsGT

variantgrid/settings/components/default_settings.py

@@ -374,6 +374,7 @@
 CLINGEN_ALLELE_REGISTRY_PASSWORD = get_secret("CLINGEN_ALLELE_REGISTRY.password")
 CLINGEN_ALLELE_REGISTRY_MAX_MANUAL_REQUESTS = 10_000  # On nodes and VCFs
 CLINGEN_ALLELE_REGISTRY_REQUIRE_REF_ALLELE = True
+CLINGEN_ALLELE_REGISTRY_REATTEMPT_WITH_ACTUAL_REF = True
 
 NO_DNA_CONTROL_REGEX = "(^|[^a-zA-Z])NDC([^a-zA-Z]|$)"  # No DNA Control - e.g. _NDC_ or -NDC_