Merge pull request #71 from griffithlab/issue_57

Downgrade duplicate variant error to a warning and skip duplicates.
griffithlab · Oct 16, 2023 · f61fb34 · f61fb34
2 parents cff9982 + 6905d5f
commit f61fb34
Show file tree

Hide file tree

Showing 4 changed files with 21 additions and 2 deletions.
diff --git a/tests/test_data/vep_annotation_reporter/input.duplicate_variant.vcf.gz b/tests/test_data/vep_annotation_reporter/input.duplicate_variant.vcf.gz
diff --git a/tests/test_data/vep_annotation_reporter/output.duplicate_variant.tsv b/tests/test_data/vep_annotation_reporter/output.duplicate_variant.tsv
@@ -0,0 +1,2 @@
+CHROM	POS	REF	ALT	SYMBOL
+chr17	7675088	C	T	TP53
diff --git a/tests/test_vep_annotation_reporter.py b/tests/test_vep_annotation_reporter.py
@@ -121,6 +121,20 @@ def test_vcf_with_multiple_transcripts_and_no_pick(self):
         self.assertTrue(cmp(os.path.join(self.test_data_dir, 'output.merge_multiple_transcripts.tsv'), os.path.join(temp_path.name, 'input.tsv')))
         temp_path.cleanup()
 
+    def test_vcf_with_duplicate_variant(self):
+        logging.disable(logging.NOTSET)
+        with LogCapture() as l:
+            temp_path = tempfile.TemporaryDirectory()
+            os.symlink(os.path.join(self.test_data_dir, 'input.duplicate_variant.vcf.gz'), os.path.join(temp_path.name, 'input.vcf.gz'))
+            command = [
+                os.path.join(temp_path.name, 'input.vcf.gz'),
+                'SYMBOL',
+            ]
+            vep_annotation_reporter.main(command)
+            self.assertTrue(cmp(os.path.join(self.test_data_dir, 'output.duplicate_variant.tsv'), os.path.join(temp_path.name, 'input.tsv')))
+            temp_path.cleanup()
+            l.check_present(('root', 'WARNING', "VEP entry at CHR chr17, POS 7675088, REF C , ALT T already exists. Skipping subsequent entries."))
+
     def test_vcf_with_multiple_transcripts_and_pick_set_for_none(self):
         logging.disable(logging.NOTSET)
         with LogCapture() as l:

diff --git a/vatools/vep_annotation_reporter.py b/vatools/vep_annotation_reporter.py
@@ -155,7 +155,7 @@ def extract_vep_fields(args):
                 else:
                     vep[chr][pos][ref][alt] = None
             else:
-                sys.exit("VEP entry for at CHR %s, POS %s, REF %s , ALT % already exists" % (chr, pos, ref, alt) )
+                logging.warning("VEP entry at CHR %s, POS %s, REF %s , ALT %s already exists. Skipping subsequent entries." % (chr, pos, ref, alt) )
     vcf_reader.close()
     return vep
 
@@ -201,6 +201,7 @@ def main(args_input = sys.argv[1:]):
         with open(output_file, 'w') as output_filehandle:
             writer = csv.DictWriter(output_filehandle, fieldnames = ['CHROM', 'POS', 'REF', 'ALT'] + args.vep_fields, delimiter = "\t")
             writer.writeheader()
+            rows = []
             for variant in vcf_reader:
                 row = {
                     'CHROM': str(variant.CHROM),
@@ -209,7 +210,9 @@ def main(args_input = sys.argv[1:]):
                     'ALT'  : ','.join(map(lambda a: a.serialize(), variant.ALT)),
                 }
                 row = add_vep_fields_to_row(args, row, vep)
-                writer.writerow(row)
+                if row not in rows:
+                    rows.append(row)
+            writer.writerows(rows)
 
 if __name__ == '__main__':
     main()