class MergeCoOccurVar(unittest.TestCase):
def setUp(self):
self.vcfio = FakeVCFIO(
{
"AF":
HeaderInfoAttr("AF", "Alternative alleles frequencies",
"Float", "A")
}, {
"AD":
HeaderFormatAttr("AD", "Alternative alleles depths", "Integer",
"A"),
"DP":
HeaderFormatAttr("DP", "total depth", "Integer", "1")
})
self.ref_seq = "ACGCAAATCTCGGCATGCCGATT"
# | | | | | | | | | |
# 1 3 5 7 9 11 14 17 20 23
self.variant_1 = VCFRecord(
"chr1", # chrom
None, # pos
"artificial_1", # id
None, # ref
None, # alt
10, # qual
["lowQual", "lowDP"], # filter
{"AF": [0.05]}, # info
["DP", "AD"], # format
{
"splA": {
"AD": [10],
"DP": 100
},
"splB": {
"AD": [40],
"DP": 4900
},
})
self.variant_2 = VCFRecord(
"chr1", # chrom
None, # pos
None, # id
None, # ref
None, # alt
30, # qual
["PASS"], # filter
{"AF": [0.06]}, # info
["DP", "AD"], # format
{
"splA": {
"AD": [5],
"DP": 50
},
"splB": {
"AD": [31],
"DP": 550
},
})
self.expected_merge = VCFRecord(
"chr1", # chrom
None, # pos
None, # id
None, # ref
None, # alt
20, # qual
["lowQual", "lowDP"], # filter
{
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=A/T", "chr1:20=G/C"]
}, # info
["DP", "AD"], # format
{
"splA": {
"AD": [5],
"DP": 50
},
"splB": {
"AD": [31],
"DP": 550
},
})
def testMergedRecord_1_substit(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "A"
self.variant_1.alt = ["T"]
# Variant 2
self.variant_2.pos = 20
self.variant_2.ref = "G"
self.variant_2.alt = ["C"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAATCTCGGCATGCCG"
self.expected_merge.alt = ["TAATCTCGGCATGCCC"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=A/T", "chr1:20=G/C"]
}
# Eval
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(), self.variant_2,
self.variant_2.getName(), self.ref_seq)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_2_largeSubstit(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "AAAT"
self.variant_1.alt = ["TGCA"]
# Variant 2
self.variant_2.pos = 10
self.variant_2.ref = "TC"
self.variant_2.alt = ["GG"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAATCTC"
self.expected_merge.alt = ["TGCACGG"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=AAAT/TGCA", "chr1:10=TC/GG"]
}
# Eval
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(), self.variant_2,
self.variant_2.getName(), self.ref_seq)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_3_largeCloseSubstit(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "AAAT"
self.variant_1.alt = ["TGCA"]
# Variant 2
self.variant_2.pos = 9
self.variant_2.ref = "CT"
self.variant_2.alt = ["GG"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAATCT"
self.expected_merge.alt = ["TGCAGG"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=AAAT/TGCA", "chr1:9=CT/GG"]
}
# Eval
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(), self.variant_2,
self.variant_2.getName(), self.ref_seq)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_4_delIns(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "AAAT"
self.variant_1.alt = ["-"]
# Variant 2
self.variant_2.pos = 10
self.variant_2.ref = "-"
self.variant_2.alt = ["GGCATCT"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAATC"
self.expected_merge.alt = ["CGGCATCT"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=AAAT/-", "chr1:10=-/GGCATCT"]
}
# Eval
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(), self.variant_2,
self.variant_2.getName(), self.ref_seq)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_5_coDelIns(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "AAAT"
self.variant_1.alt = ["-"]
# Variant 2
self.variant_2.pos = 9
self.variant_2.ref = "-"
self.variant_2.alt = ["AGG"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAAT"
self.expected_merge.alt = ["AGG"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=AAAT/-", "chr1:9=-/AGG"]
}
# Eval
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(), self.variant_2,
self.variant_2.getName(), self.ref_seq)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_6_insDel(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "-"
self.variant_1.alt = ["GTGTG"]
# Variant 2
self.variant_2.pos = 7
self.variant_2.ref = "ATC"
self.variant_2.alt = ["-"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAATC"
self.expected_merge.alt = ["GTGTGAA"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=-/GTGTG", "chr1:7=ATC/-"]
}
# Eval
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(), self.variant_2,
self.variant_2.getName(), self.ref_seq)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_7_closeInsDel(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "-"
self.variant_1.alt = ["GTGTG"]
# Variant 2
self.variant_2.pos = 6
self.variant_2.ref = "AA"
self.variant_2.alt = ["-"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAA"
self.expected_merge.alt = ["GTGTGA"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=-/GTGTG", "chr1:6=AA/-"]
}
# Eval
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(), self.variant_2,
self.variant_2.getName(), self.ref_seq)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_8_coInsDel(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "-"
self.variant_1.alt = ["GTGTG"]
# Variant 2
self.variant_2.pos = 5
self.variant_2.ref = "AA"
self.variant_2.alt = ["-"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AA"
self.expected_merge.alt = ["GTGTG"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=-/GTGTG", "chr1:5=AA/-"]
}
# Eval
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(), self.variant_2,
self.variant_2.getName(), self.ref_seq)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
class MergeCoOccurVar(unittest.TestCase):
def setUp(self):
# VCF
self.vcfio = FakeVCFIO(
{
"AF":
HeaderInfoAttr("AF", "Alternative alleles frequencies",
"Float", "A")
}, {
"AD":
HeaderFormatAttr("AD", "Alternative alleles depths", "Integer",
"A"),
"DP":
HeaderFormatAttr("DP", "total depth", "Integer", "1")
})
# Ref seq
tmp_folder = tempfile.gettempdir()
unique_id = str(uuid.uuid1())
self.tmp_fasta_path = os.path.join(tmp_folder, unique_id + ".fa")
self.tmp_faidx_path = os.path.join(tmp_folder, unique_id + ".fa.fai")
self.ref_seq = "ACGCAAATCTCGGCATGCCGATT"
# | | | | | | | | | |
# 1 3 5 7 9 11 14 17 20 23
with open(self.tmp_fasta_path, "w") as FH_seq:
FH_seq.write(">chr1\n{}".format(self.ref_seq))
with open(self.tmp_faidx_path, "w") as FH_faidx:
FH_faidx.write("chr1\t{}\t6\t60\t61".format(len(self.ref_seq)))
# Variants
self.variant_1 = VCFRecord(
"chr1", # chrom
None, # pos
"artificial_1", # id
None, # ref
None, # alt
10, # qual
["lowQual", "lowDP"], # filter
{"AF": [0.05]}, # info
["DP", "AD"], # format
{
"splA": {
"AD": [10],
"DP": 100
},
"splB": {
"AD": [40],
"DP": 4900
},
})
self.variant_2 = VCFRecord(
"chr1", # chrom
None, # pos
None, # id
None, # ref
None, # alt
30, # qual
["PASS"], # filter
{"AF": [0.06]}, # info
["DP", "AD"], # format
{
"splA": {
"AD": [5],
"DP": 50
},
"splB": {
"AD": [31],
"DP": 550
},
})
self.expected_merge = VCFRecord(
"chr1", # chrom
None, # pos
None, # id
None, # ref
None, # alt
20, # qual
["lowQual", "lowDP"], # filter
{
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=A/T", "chr1:20=G/C"]
}, # info
["DP", "AD"], # format
{
"splA": {
"AD": [5],
"DP": 50
},
"splB": {
"AD": [31],
"DP": 550
},
})
def tearDown(self):
# Clean temporary files
for curr_file in [self.tmp_fasta_path, self.tmp_faidx_path]:
if os.path.exists(curr_file):
os.remove(curr_file)
def testMergedRecord_1_substit(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "A"
self.variant_1.alt = ["T"]
# Variant 2
self.variant_2.pos = 20
self.variant_2.ref = "G"
self.variant_2.alt = ["C"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAATCTCGGCATGCCG"
self.expected_merge.alt = ["TAATCTCGGCATGCCC"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=A/T", "chr1:20=G/C"]
}
# Eval
with IdxFastaIO(self.tmp_fasta_path) as FH_ref:
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(),
self.variant_2,
self.variant_2.getName(), FH_ref)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_2_largeSubstit(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "AAAT"
self.variant_1.alt = ["TGCA"]
# Variant 2
self.variant_2.pos = 10
self.variant_2.ref = "TC"
self.variant_2.alt = ["GG"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAATCTC"
self.expected_merge.alt = ["TGCACGG"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=AAAT/TGCA", "chr1:10=TC/GG"]
}
# Eval
with IdxFastaIO(self.tmp_fasta_path) as FH_ref:
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(),
self.variant_2,
self.variant_2.getName(), FH_ref)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_3_largeCloseSubstit(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "AAAT"
self.variant_1.alt = ["TGCA"]
# Variant 2
self.variant_2.pos = 9
self.variant_2.ref = "CT"
self.variant_2.alt = ["GG"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAATCT"
self.expected_merge.alt = ["TGCAGG"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=AAAT/TGCA", "chr1:9=CT/GG"]
}
# Eval
with IdxFastaIO(self.tmp_fasta_path) as FH_ref:
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(),
self.variant_2,
self.variant_2.getName(), FH_ref)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_4_delIns(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "AAAT"
self.variant_1.alt = ["-"]
# Variant 2
self.variant_2.pos = 10
self.variant_2.ref = "-"
self.variant_2.alt = ["GGCATCT"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAATC"
self.expected_merge.alt = ["CGGCATCT"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=AAAT/-", "chr1:10=-/GGCATCT"]
}
# Eval
with IdxFastaIO(self.tmp_fasta_path) as FH_ref:
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(),
self.variant_2,
self.variant_2.getName(), FH_ref)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_5_coDelIns(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "AAAT"
self.variant_1.alt = ["-"]
# Variant 2
self.variant_2.pos = 9
self.variant_2.ref = "-"
self.variant_2.alt = ["AGG"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAAT"
self.expected_merge.alt = ["AGG"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=AAAT/-", "chr1:9=-/AGG"]
}
# Eval
with IdxFastaIO(self.tmp_fasta_path) as FH_ref:
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(),
self.variant_2,
self.variant_2.getName(), FH_ref)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_6_insDel(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "-"
self.variant_1.alt = ["GTGTG"]
# Variant 2
self.variant_2.pos = 7
self.variant_2.ref = "ATC"
self.variant_2.alt = ["-"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAATC"
self.expected_merge.alt = ["GTGTGAA"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=-/GTGTG", "chr1:7=ATC/-"]
}
# Eval
with IdxFastaIO(self.tmp_fasta_path) as FH_ref:
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(),
self.variant_2,
self.variant_2.getName(), FH_ref)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_7_closeInsDel(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "-"
self.variant_1.alt = ["GTGTG"]
# Variant 2
self.variant_2.pos = 6
self.variant_2.ref = "AA"
self.variant_2.alt = ["-"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AAA"
self.expected_merge.alt = ["GTGTGA"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=-/GTGTG", "chr1:6=AA/-"]
}
# Eval
with IdxFastaIO(self.tmp_fasta_path) as FH_ref:
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(),
self.variant_2,
self.variant_2.getName(), FH_ref)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))
def testMergedRecord_8_coInsDel(self):
# Variant 1
self.variant_1.pos = 5
self.variant_1.ref = "-"
self.variant_1.alt = ["GTGTG"]
# Variant 2
self.variant_2.pos = 5
self.variant_2.ref = "AA"
self.variant_2.alt = ["-"]
# Expected merge
self.expected_merge.pos = 5
self.expected_merge.ref = "AA"
self.expected_merge.alt = ["GTGTG"]
self.expected_merge.info = {
"AF": [0.06],
"MCO_QUAL": [10, 30],
"MCO_VAR": ["chr1:5=-/GTGTG", "chr1:5=AA/-"]
}
# Eval
with IdxFastaIO(self.tmp_fasta_path) as FH_ref:
observed_merge = mergedRecord(self.vcfio, self.variant_1,
self.variant_1.getName(),
self.variant_2,
self.variant_2.getName(), FH_ref)
self.assertEqual(strVariant(observed_merge),
strVariant(self.expected_merge))