Merge pull request #675 from openvar/develop

Develop

Author: Peter-J-Freeman

VariantValidator/modules/complex_descriptions.py

@@ -281,6 +281,15 @@ def uncertain_positions(my_variant, validator):
     if not re.search("[gcnr].\(", my_variant.quibble):
         return
 
+    # Check for LRGs
+    hgvs_accession, variation = my_variant.quibble.split(":")
+    if "LRG" in hgvs_accession:
+        if "t" in hgvs_accession:
+            hgvs_accession = validator.db.get_refseq_transcript_id_from_lrg_transcript_id(hgvs_accession)
+        else:
+            hgvs_accession = validator.db.get_refseq_id_from_lrg_id(hgvs_accession)
+        my_variant.quibble = f"{hgvs_accession}:{variation}"
+
     # Formats like NC_000005.9:g.(90136803_90144453)_(90159675_90261231)dup
     if ")_(" in my_variant.quibble and "?" not in my_variant.quibble:
         accession, _sep, var_type_and_posedit = my_variant.quibble.partition(':')
@@ -311,6 +320,7 @@ def uncertain_positions(my_variant, validator):
                     end=end)
         except vvhgvs.exceptions.HGVSError as e:
             raise HgvsParseError(str(e))
+
         try:
             validator.vr.validate(parsed_v1)
         except vvhgvs.exceptions.HGVSError as e:

VariantValidator/modules/format_converters.py

@@ -812,6 +812,13 @@ def intronic_converter(variant, validator, skip_check=False, uncertain=False):
                 pass
             elif "does not agree with reference sequence" in str(e):
                 previous_exception = e
+                try:
+                    check_g = validator.vm.t_to_g(hgvs_transy, hgvs_genomic.ac,
+                                                  alt_aln_method=validator.alt_aln_method)
+                    validator.vm.g_to_t(check_g, hgvs_transy.ac, alt_aln_method=validator.alt_aln_method)
+                except vvhgvs.exceptions.HGVSError as e:
+                    if "start or end or both are beyond the bounds of transcript record" in str(e):
+                        raise
                 try:
                     variant.hn.normalize(hgvs_transy)
                 except vvhgvs.exceptions.HGVSUnsupportedOperationError as e:
@@ -834,8 +841,8 @@ def intronic_converter(variant, validator, skip_check=False, uncertain=False):
                 else:
                     variant.warnings.append(f'ExonBoundaryError: {hgvs_transy.posedit.pos} does not match the exon '
                                             f'boundaries for the alignment of {hgvs_transy.ac} to {hgvs_genomic.ac}')
-            except vvhgvs.exceptions.HGVSError:
-                pass
+            except vvhgvs.exceptions.HGVSError as e:
+               pass
 
     logger.debug("HVGS typesetting complete")
 

VariantValidator/modules/vvMixinCore.py

@@ -341,8 +341,8 @@ def validate(self,
                                                                               select_transcripts_dict_plus_version)
 
                     except vvhgvs.exceptions.HGVSError as e:
-                        import traceback
-                        traceback.print_exc()
+                        # import traceback
+                        # traceback.print_exc()
                         checkref = str(e)
                         try:
                             # Test intronic variants for incorrect boundaries (see issue #169)
@@ -1081,8 +1081,12 @@ def validate(self,
                     variant.gene_symbol = self.db.get_gene_symbol_from_transcript_id(
                         variant.hgvs_transcript_variant.ac)
                 elif variant.hgvs_refseqgene_variant and variant.gene_symbol == '':
-                    variant.gene_symbol = self.db.get_gene_symbol_from_refseq_id(
-                        variant.hgvs_refseqgene_variant.ac)
+                    try:
+                        variant.gene_symbol = self.db.get_gene_symbol_from_refseq_id(
+                            variant.hgvs_refseqgene_variant.ac)
+                    except AttributeError:
+                        variant.gene_symbol = self.db.get_gene_symbol_from_refseq_id(
+                            variant.hgvs_refseqgene_variant.split(":")[0])
 
                 # Add stable gene_ids
                 stable_gene_ids = {}

tests/test_warnings.py

@@ -166,6 +166,13 @@ def test_issue_338h(self):
         assert 'because no reference sequence ID has been provided' in \
                results['NM_000088.3:c.589G>T']['validation_warnings'][0]
 
+    def test_issue_673a(self):
+        variant = 'NC_000001.10(NM_001128425.1):c.35+11000000C>T'
+        results = self.vv.validate(variant, 'GRCh38', 'all', liftover_level='primary').format_as_dict(test=True)
+        print(results)
+        assert 'start or end or both are beyond the bounds of transcript record' in \
+               results['validation_warning_1']['validation_warnings']
+
     def test_issue_359(self):
         variant = 'NM_001371623.1:c.483ins'
         results = self.vv.validate(variant, 'GRCh37', 'all', liftover_level='primary').format_as_dict(test=True)
@@ -535,6 +542,13 @@ def test_uncertain_5(self):
         assert "Position 90159675_90261231 is > or overlaps 90136803_90144453" in results[
             'validation_warning_1']["validation_warnings"]
 
+    def test_uncertain_6(self):
+        variant = 'LRG_199t1:c.(6278_6293-1)_(7542+49_757000005)dup'
+        results = self.vv.validate(variant, 'GRCh38', 'all').format_as_dict(test=True)
+        print(results)
+        assert "Variant coordinate is out of the bound of CDS region (CDS length : 11058)" in results[
+            'validation_warning_1']["validation_warnings"]
+
     def test_uncertain_fuzzy_1(self):
         variant = 'NC_000002.12:g.(?_187315204)_(192885646_?)del'
         results = self.vv.validate(variant, 'GRCh38', 'all').format_as_dict(test=True)