Handle cases where stop codon is X, not *

lib/fasta_generator.py

@@ -106,9 +106,14 @@ def execute(self):
                 wildtype_amino_acid, mutant_amino_acid = line['amino_acid_change'].split('/')
                 if '*' in wildtype_amino_acid:
                     wildtype_amino_acid = wildtype_amino_acid.split('*')[0]
+                elif 'X' in wildtype_amino_acid:
+                    wildtype_amino_acid = wildtype_amino_acid.split('X')[0]
                 if '*' in mutant_amino_acid:
                     mutant_amino_acid = mutant_amino_acid.split('*')[0]
                     stop_codon_added = True
+                elif 'X' in mutant_amino_acid:
+                    mutant_amino_acid = mutant_amino_acid.split('X')[0]
+                    stop_codon_added = True
                 else:
                     stop_codon_added = False
                 if wildtype_amino_acid == '-':
@@ -126,9 +131,14 @@ def execute(self):
                 wildtype_amino_acid, mutant_amino_acid = line['amino_acid_change'].split('/')
                 if '*' in wildtype_amino_acid:
                     wildtype_amino_acid = wildtype_amino_acid.split('*')[0]
+                elif 'X' in wildtype_amino_acid:
+                    wildtype_amino_acid = wildtype_amino_acid.split('X')[0]
                 if '*' in mutant_amino_acid:
                     mutant_amino_acid = mutant_amino_acid.split('*')[0]
                     stop_codon_added = True
+                elif 'X' in mutant_amino_acid:
+                    mutant_amino_acid = mutant_amino_acid.split('X')[0]
+                    stop_codon_added = True
                 else:
                     stop_codon_added = False
                 position = int(line['protein_position'].split('-', 1)[0]) - 1

tests/test_data/fasta_generator/protein_change_with_X_in_wildtype_and_mutant/input.tsv

@@ -0,0 +1,2 @@
+chromosome_name	start	stop	reference	variant	gene_name	transcript_name	amino_acid_change	ensembl_gene_id	wildtype_amino_acid_sequence	downstream_amino_acid_sequence	fusion_amino_acid_sequence	variant_type	protein_position	transcript_expression	gene_expression	normal_depth	normal_vaf	tdna_depth	tdna_vaf	trna_depth	trna_vaf	index	protein_length_change
+12	96617457	96617460	CAGA	C	ELK3	ENST00000547249	AX/X	ENSG00000111145	MESAITLWQFLLQLLLDQKHEHLICWTSNDGEFKLLKA			inframe_del	38-39	NA	NA	NA	NA	NA	NA	NA	NA	39805.ELK3.ENST00000547249.inframe_del.38-39AX/X	

tests/test_fasta_generator.py

@@ -564,6 +564,27 @@ def test_protein_change_with_asterisk_in_wildtype_and_mutant(self):
         self.assertEqual(os.path.getsize(generate_fasta_output_file.name), 0)
         self.assertEqual(os.path.getsize(generate_fasta_key_output_file.name), 0)
 
+    def test_protein_change_with_X_in_wildtype_and_mutatnt(self):
+        peptide_sequence_length        = '21'
+        test_data_dir                  = os.path.join(self.test_data_dir, 'protein_change_with_X_in_wildtype_and_mutant')
+        generate_fasta_input_file      = os.path.join(test_data_dir, 'input.tsv')
+        generate_fasta_output_file     = tempfile.NamedTemporaryFile()
+        generate_fasta_key_output_file = tempfile.NamedTemporaryFile()
+
+        generate_fasta_params = {
+            'input_file'                : generate_fasta_input_file,
+            'peptide_sequence_length'   : self.peptide_sequence_length,
+            'epitope_length'            : self.epitope_length,
+            'output_file'               : generate_fasta_output_file.name,
+            'output_key_file'           : generate_fasta_key_output_file.name,
+            'downstream_sequence_length': None,
+        }
+        generator = FastaGenerator(**generate_fasta_params)
+
+        self.assertFalse(generator.execute())
+        self.assertEqual(os.path.getsize(generate_fasta_output_file.name), 0)
+        self.assertEqual(os.path.getsize(generate_fasta_key_output_file.name), 0)
+
     def test_protein_change_with_multiple_asterisks(self):
         peptide_sequence_length        = '21'
         test_data_dir                  = os.path.join(self.test_data_dir, 'protein_change_with_multiple_asterisks')