def setUp(self):
self.vcf_file1 = [
'\t'.join(['chr1', '16688', '.', 'G', 'A', '.', '', '', '', '', '']),
'\t'.join(['chr1', '186478', '.', 'A', 'G', '.', '', '', '', '', '']),
'\t'.join(['chr1', '16577291', '.', 'T', 'C', '.', '', '', '', '', ''])]
self.vcf1 = parse_vcf_file(self.vcf_file1)
self.vcf2 = parse_vcf_file(self.vcf_file1)
def test_remove_duplicates_at_same_loc(self):
vcf_loc_dup = ['\t'.join(['chr1', '186478', '.', 'T', 'G', '.', '', '', '', '', ''])]
vcf = parse_vcf_file(vcf_loc_dup) + self.vcf1
vcf_rem_true_dup = vcf.remove_true_duplicates()
vcf_rem_loc_dup = vcf.remove_loc_dup()
self.assertEqual(len(vcf), len(self.vcf1) + len(vcf_loc_dup))
self.assertEqual(len(vcf_rem_true_dup), len(vcf))
self.assertEqual(vcf_rem_true_dup, vcf)
self.assertEqual(len(vcf_rem_loc_dup), len(self.vcf1))
def merge_vcfs(ctx,
vcf_file1,
vcf_file2,
vcf_out,
ann_vcf1=None,
ann_vcf2=None):
logging.info(f'Reading file: {vcf_file1}')
vcf1 = parse_vcf_file(read_vcf(vcf_file1))
logging.info(f'Reading file: {vcf_file2}')
vcf2 = parse_vcf_file(read_vcf(vcf_file2))
if ann_vcf1 is not None:
vcf1 = vcf1.annotate(ann_vcf1)
if ann_vcf2 is not None:
vcf2 = vcf2.annotate(ann_vcf2)
logging.info('Merging VCFs...')
vcf = vcf1 + vcf2
n1 = len(vcf)
vcf = vcf.remove_true_duplicates()
n2 = len(vcf)
logging.info(f'{n1-n2} duplicates removed')
vcf.to_file(vcf_out)
def test_merge_two_vcfs(self):
vcf = self.vcf1 + self.vcf2
self.assertEqual(vcf, parse_vcf_file(self.vcf_file1 * 2))
def test_parse_vcf_file_with_header(self):
header = ['# This is a header', '# Multiline header here']
vcf = parse_vcf_file(header + self.vcf_file)
self.assertEqual(len(vcf), len(self.vcf_file))
self.assertEqual(len(vcf.header), len(header))
def test_parse_vcf_file(self):
vcf = parse_vcf_file(self.vcf_file)
self.assertIsInstance(vcf, VCF)
self.assertEqual(len(vcf), len(self.vcf_file))
self.assertEqual(vcf[0], _parse_vcf_line(self.vcf_file[0]))