Accept transcript accession number as transcript variant identifier.

* Internally the accession number is translated to the transcript name
  (i.e. v-number) and subsequent processing is untouched.

Added failing test with accession number as transcript variant identifier.

Fixed comments by @martijnvermaat in PR #405

mutalyzer/GenRecord.py

@@ -312,6 +312,21 @@ def findGene(self, name) :
         return None
     #findGene
 
+    def get_transcript_selector(self, accession):
+        """
+        Returns a tuple with gene name and transcript name (i.e. its
+        `v-number') for a given transcript ID.
+
+        @param accession: unicode
+        @return: tuple(unicode, unicode)
+        """
+
+        for gene in self.geneList:
+            for transcript in gene.transcriptList:
+                if transcript.transcriptID == accession:
+                    return gene.name, transcript.name
+    #getInfoByTranscriptID
+
     def listGenes(self) :
         """
         List the names of all genes found in this record.

mutalyzer/grammar.py

@@ -83,9 +83,19 @@ class Grammar():
     # BNF: GeneName -> ([a-Z] | [0-9] | `-')+
     GeneName = Word(unicode(alphanums) + '-', min=1)
 
-    # BNF: GeneSymbol -> `(' Name (TransVar | ProtIso)? `)'
-    GeneSymbol = Suppress('(') + Group(GeneName('GeneSymbol') + \
-                 Optional(TransVar ^ ProtIso))('Gene') + Suppress(')')
+    # BNF: GeneProductID -> GeneName (TransVar | ProtIso)
+    GeneProductID = Group(GeneName('GeneSymbol') + \
+                          Optional(TransVar ^ ProtIso))('Gene')
+
+    # BNF: AccNoStem -> ([a-Z] Number `_')+
+    AccNoStem = NotAny('LRG_') + Combine(Word(unicode(alphas) + '_') + Number)
+
+    # BNF: AccNoFull -> AccNoStem `.' Number
+    AccNoFull = AccNoStem + Suppress('.') + Number
+
+    # BNF: GeneSymbol -> `(' (GeneProductID | AccNoFull) `)'
+    GeneSymbol = Suppress('(') + (GeneProductID ^ AccNoFull('AccNoTransVar')) \
+                 + Suppress(')')
 
     # BNF: GI -> (`GI' | `GI:')? Number
     GI = Suppress(Optional('GI') ^ Optional('GI:') ^ Optional('gi') ^

mutalyzer/variantchecker.py

@@ -1751,6 +1751,19 @@ def check_variant(description, output):
     if not retrieved_record:
         return
 
+    if gene_symbol == '' and transcript_id == '' and \
+        parsed_description.AccNoTransVar:
+        # Get gene symbol and transcript name by transcript accession number.
+        transcript_info = retrieved_record.get_transcript_selector(
+            '.'.join(parsed_description.AccNoTransVar))
+        if transcript_info is not None:
+            gene_symbol, transcript_id = transcript_info
+        else:
+            output.addMessage(__file__, 4, 'EINVALIDTRANSVAR',
+                'Invalid name for transcript variant identifier.')
+            return
+
+
     # Add recordType to output for output formatting.
     output.addOutput('recordType', filetype)
     output.addOutput('organism', retrieved_record.organism)

tests/test_variantchecker.py

@@ -2214,3 +2214,23 @@ def test_dup_range_arg_seq_reverse(output, checker):
     assert len(e_nodna) == 0
     e_ref = output.getMessagesWithErrorCode('EREF')
     assert len(e_ref) == 0
+
+
+@with_references('NG_012337.1')
+def test_accno_as_transcript_variant(output, checker):
+    """
+    Test accession number as transcript variant identifier is accepted.
+    """
+    # Variant below should be equivalent to the already accepted
+    # version: 'NG_012337.1(TIMM8B_v001):c.12_13insGATC'
+    checker('NG_012337.1(NM_012459.2):c.12_13insGATC')
+
+    errorcount, warncount, summary = output.Summary()
+    assert errorcount == 0
+    assert output.getOutput('gDescription')[0] == u'g.4911_4912insATCG'
+
+    # Check non-existing accession number as transcript variant identifier.
+    checker('NG_012337.1(DUMMYACCNO_9999.9):c.12_13insGATC')
+
+    e_invalidtransvar = output.getMessagesWithErrorCode('EINVALIDTRANSVAR')
+    assert len(e_invalidtransvar) == 1