def test_003_check_merge_multi(self):
ref_seq = 'ATGGTATGCGATTGACC'
chrom='chrom1'
h1 = [Variant(chrom, 0, 'ATG', alt='G', qual=5, genotype_data={'GT':'1/1'}),
Variant(chrom, 4, 'T', alt='G', qual=5, genotype_data={'GT':'1/1'}),
Variant(chrom, 7, 'G', alt='GG', qual=5, genotype_data={'GT':'1/1'}),
Variant(chrom, 9, ref_seq[9], alt=ref_seq[9] + 'T', qual=5, genotype_data={'GT':'1/1'}),
]
h2 = [Variant(chrom, 1, 'T', alt='TT', qual=10, genotype_data={'GT':'1/1'}),
Variant(chrom, 2, ref_seq[2:10], alt=ref_seq[2], qual=10, genotype_data={'GT':'1/1'}),
]
# POS 0 1 2 3 4 5 6 7 8 9 10
# REF A T G G T A T G C G A
# H1 - - G G g A T Gg C Gt A
# H2 A Tt G - - - - - - - A
# expected merged variants
ref_expt = "ATGGTATGCG"
alt1_expt = "GGGATGGCGT"
alt2_expt = "ATTG"
expt = Variant(chrom, 0, ref_expt, alt=[alt1_expt, alt2_expt], qual=7.5, genotype_data={'GT': '1|2'})
comb_interval, trees = self.intervaltree_prep(h1, h2, ref_seq)
# preserve phase otherwise alts could be switched around
got = _merge_variants(comb_interval, trees, ref_seq, discard_phase=False)
for key in ('chrom', 'pos', 'ref', 'qual', 'alt', 'gt', 'phased'):
expected = getattr(expt, key)
result = getattr(got, key)
self.assertEqual(expected, result, 'Merging failed for {}:{} {}.'.format(expt.chrom, expt.pos+1, key))
def test_002_check_merge_snps(self):
ref_seq = 'ATGGTATGCGATTGACC'
chrom = 'chrom1'
h1 = [
Variant(chrom, 0, 'A', alt='C', qual=5, sample_dict={'GT': '1/1'})
]
h2 = [
Variant(chrom, 0, 'A', alt='T', qual=10, sample_dict={'GT': '1/1'})
]
expt = Variant(chrom,
0,
'A',
alt=['C', 'T'],
qual=7.5,
sample_dict={'GT': '1/2'})
comb_interval, trees = self.intervaltree_prep(h1, h2, ref_seq)
# preserve phase otherwise alts could be switched around
got = _merge_variants(comb_interval,
trees,
ref_seq,
discard_phase=False)
for key in ('chrom', 'pos', 'ref', 'qual', 'alt', 'gt'):
expected = getattr(expt, key)
result = getattr(got, key)
self.assertEqual(
expected, result,
'Merging failed for {}:{} {}.'.format(expt.chrom, expt.pos + 1,
key))
def test_004_check_merge_multi_bug(self):
# if we have two indels on one haplotype that cancel each other out
# (e.g. insertion of a T followed by a deletion of a T)
# check we don't have an alt that is the same as the ref.
ref_seq = 'TTTTTTTTTT'
chrom = 'chrom1'
h1 = [
Variant(chrom,
0,
'TTTTT',
alt='T',
qual=5,
genotype_data={'GT': '1/1'})
]
h2 = [
Variant(chrom,
1,
'T',
alt='TT',
qual=10,
genotype_data={'GT': '1/1'}),
Variant(chrom,
3,
'TT',
alt='T',
qual=10,
genotype_data={'GT': '1/1'})
]
expt = Variant(chrom,
0,
'TTTTT',
alt='T',
qual=5,
genotype_data={'GT': '1|0'})
comb_interval, trees = self.intervaltree_prep(h1, h2, ref_seq)
# preserve phase otherwise alts could be switched around
got = _merge_variants(comb_interval,
trees,
ref_seq,
discard_phase=False)
for key in ('chrom', 'pos', 'ref', 'qual', 'alt', 'gt', 'phased'):
expected = getattr(expt, key)
result = getattr(got, key)
self.assertEqual(
expected, result,
'Merging failed for {}:{} {}.'.format(expt.chrom, expt.pos + 1,
key))
