def testUnsupportedCigarToken(self):
cigar = '3M3X6M'
seq = 'AAACAACCACCC' # @IgnorePep8
quality = 'BBBDDDEEEFFF'
with self.assertRaises(RuntimeError) as result:
apply_cigar(cigar, seq, quality)
self.assertEqual("Unsupported CIGAR token: '3X'.",
result.exception.message)
def testInvalidCigar(self):
cigar = '3M...6M'
seq = 'AAACAACCACCC' # @IgnorePep8
quality = 'BBBDDDEEEFFF'
with self.assertRaises(RuntimeError) as result:
apply_cigar(cigar, seq, quality)
self.assertEqual("Invalid CIGAR string: '3M...6M'.",
result.exception.message)
def testLongCigar(self):
cigar = '10M'
seq = 'AAACAACCA' # @IgnorePep8
quality = 'BBBDDDEEE'
with self.assertRaises(RuntimeError) as result:
apply_cigar(cigar, seq, quality)
self.assertEqual(
"CIGAR string '10M' is too long for sequence 'AAACAACCA'.",
result.exception.message)
def testUnsupportedCigarToken(self):
cigar = '3M3X6M'
seq = 'AAACAACCACCC' # @IgnorePep8
quality = 'BBBDDDEEEFFF'
with self.assertRaises(RuntimeError) as result:
apply_cigar(cigar, seq, quality)
self.assertEqual(
"Unsupported CIGAR token: '3X'.",
result.exception.message)
def testLongCigar(self):
cigar = '10M'
seq = 'AAACAACCA' # @IgnorePep8
quality = 'BBBDDDEEE'
with self.assertRaises(RuntimeError) as result:
apply_cigar(cigar, seq, quality)
self.assertEqual(
"CIGAR string '10M' is too long for sequence 'AAACAACCA'.",
result.exception.message)
def testInvalidCigar(self):
cigar = '3M...6M'
inp_sequence = 'AAACAACCACCC'
inp__quality = 'BBBDDDEEEFFF'
with self.assertRaises(RuntimeError) as result:
apply_cigar(cigar, inp_sequence, inp__quality)
self.assertEqual(
"Invalid CIGAR string: '3M...6M'.",
result.exception.args[0])
def testUnsupportedCigarToken(self):
cigar = '3M3X6M'
inp_sequence = 'AAACAACCACCC'
inp__quality = 'BBBDDDEEEFFF'
with self.assertRaises(RuntimeError) as result:
apply_cigar(cigar, inp_sequence, inp__quality)
self.assertEqual(
"Unsupported CIGAR token: '3X'.",
result.exception.args[0])
def testLongCigar(self):
cigar = '10M'
inp_sequence = 'AAACAACCA'
inp__quality = 'BBBDDDEEE'
with self.assertRaises(RuntimeError) as result:
apply_cigar(cigar, inp_sequence, inp__quality)
self.assertEqual(
"CIGAR string '10M' is too long for sequence 'AAACAACCA'.",
result.exception.args[0])
def testInvalidCigar(self):
cigar = '3M...6M'
seq = 'AAACAACCACCC' # @IgnorePep8
quality = 'BBBDDDEEEFFF'
with self.assertRaises(RuntimeError) as result:
apply_cigar(cigar, seq, quality)
self.assertEqual(
"Invalid CIGAR string: '3M...6M'.",
result.exception.message)
def merge_reads(quality_cutoff, read_pair):
""" Merge a pair of reads.
Also skip reads that don't meet certain criteria.
@param quality_cutoff: minimum quality score for a base to be counted
@param read_pair: a sequence of two sequences, each with fields from a
SAM file record
@return: (rname, mseq, merged_inserts, qual1, qual2) or None to skip the pair
"""
read1, read2 = read_pair
if read2 and read1[2] != read2[2]:
# region mismatch, ignore the read pair.
return None
filtered_reads = []
rname = None
for read in read_pair:
if not read:
continue
(_qname,
flag,
rname,
refpos_str,
_mapq,
cigar,
_rnext,
_pnext,
_tlen,
seq,
qual) = read[:11] # ignore optional fields
if is_unmapped_read(flag):
continue
filtered_reads.append(dict(rname=rname,
cigar=cigar,
seq=seq,
qual=qual,
pos=int(refpos_str)))
if not filtered_reads:
return None
seq1, qual1, ins1 = apply_cigar(filtered_reads[0]['cigar'],
filtered_reads[0]['seq'],
filtered_reads[0]['qual'],
filtered_reads[0]['pos']-1)
if len(filtered_reads) == 1:
seq2 = qual2 = ''
ins2 = None
else:
seq2, qual2, ins2 = apply_cigar(filtered_reads[1]['cigar'],
filtered_reads[1]['seq'],
filtered_reads[1]['qual'],
filtered_reads[1]['pos']-1)
mseq = merge_pairs(seq1, seq2, qual1, qual2, q_cutoff=quality_cutoff)
merged_inserts = merge_inserts(ins1, ins2, quality_cutoff)
return rname, mseq, merged_inserts, qual1, qual2
def merge_reads(quality_cutoff, read_pair):
""" Merge a pair of reads.
Also skip reads that don't meet certain criteria.
@param quality_cutoff: minimum quality score for a base to be counted
@param read_pair: a sequence of two sequences, each with fields from a
SAM file record
@return: (rname, mseq, merged_inserts, qual1, qual2) or None to skip the pair
"""
read1, read2 = read_pair
if read2 and read1[2] != read2[2]:
# region mismatch, ignore the read pair.
return None
filtered_reads = []
for read in read_pair:
if not read:
continue
(_qname,
flag,
rname,
refpos_str,
_mapq,
cigar,
_rnext,
_pnext,
_tlen,
seq,
qual) = read[:11] # ignore optional fields
if is_unmapped_read(flag):
continue
filtered_reads.append(dict(rname=rname,
cigar=cigar,
seq=seq,
qual=qual,
pos=int(refpos_str)))
if not filtered_reads:
return None
seq1, qual1, ins1 = apply_cigar(filtered_reads[0]['cigar'],
filtered_reads[0]['seq'],
filtered_reads[0]['qual'],
filtered_reads[0]['pos']-1)
if len(filtered_reads) == 1:
seq2 = qual2 = ''
ins2 = None
else:
seq2, qual2, ins2 = apply_cigar(filtered_reads[1]['cigar'],
filtered_reads[1]['seq'],
filtered_reads[1]['qual'],
filtered_reads[1]['pos']-1)
mseq = merge_pairs(seq1, seq2, qual1, qual2, q_cutoff=quality_cutoff)
merged_inserts = merge_inserts(ins1, ins2, quality_cutoff)
return rname, mseq, merged_inserts, qual1, qual2
def testSoftClipPositions(self):
cigar = '3S6M'
pos = 4
inp__sequence = 'AAACAACCA'
inp___quality = 'BBBDDDEEE'
exp_sequence = '----CAACCA'
exp__quality = '!!!!DDDEEE'
mapped = set()
soft_clipped = set()
expected_mapped = {4, 5, 6, 7, 8, 9}
expected_soft_clipped = {1, 2, 3}
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar,
inp__sequence,
inp___quality,
pos=pos,
mapped=mapped,
soft_clipped=soft_clipped)
self.assertEqual(exp_sequence, clipped_seq)
self.assertEqual(exp__quality, clipped_quality)
self.assertEqual({}, inserts)
self.assertEqual(expected_mapped, mapped)
self.assertEqual(expected_soft_clipped, soft_clipped)
def testLargeToken(self):
cigar = '12M'
seq = 'AAACAACCACCC' # @IgnorePep8
quality = 'BBBBBBBBBBBB'
expected_seq = seq
expected_quality = quality
clipped_seq, clipped_quality, inserts = apply_cigar(cigar, seq, quality)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testInsertionLowQuality(self):
cigar = '3M3I6M'
seq = 'AAACAACCACCC' # @IgnorePep8
quality = 'BBBD*DEEEFFF' # @IgnorePep8
expected_seq = 'AAACCACCC' # @IgnorePep8
expected_quality = 'BBBEEEFFF'
clipped_seq, clipped_quality, inserts = apply_cigar(cigar, seq, quality)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({3: ('CAA', 'D*D')}, inserts)
def testSoftClip(self):
cigar = '3S6M'
seq = 'AAACAACCA' # @IgnorePep8
quality = 'BBBDDDEEE' # @IgnorePep8
expected_seq = 'CAACCA' # @IgnorePep8
expected_quality = 'DDDEEE' # @IgnorePep8
clipped_seq, clipped_quality, inserts = apply_cigar(cigar, seq, quality)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testDeletion(self):
cigar = '6M3D3M'
seq = 'AAACAACCA' # @IgnorePep8
quality = 'BBBDDDEEE' # @IgnorePep8
expected_seq = 'AAACAA---CCA' # @IgnorePep8
expected_quality = 'BBBDDD EEE'
clipped_seq, clipped_quality, inserts = apply_cigar(cigar, seq, quality)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testTrivial(self):
cigar = '9M'
seq = 'AAACAACCA' # @IgnorePep8
quality = 'BBBBBBBBB'
expected_seq = seq
expected_quality = quality
clipped_seq, clipped_quality, inserts = apply_cigar(cigar, seq, quality)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testDeletion(self):
cigar = '6M3D3M'
inp_sequence = 'AAACAACCA'
inp__quality = 'BBBDDDEEE'
exp_sequence = 'AAACAA---CCA'
exp__quality = 'BBBDDD EEE'
clipped_seq, clipped_quality, inserts = apply_cigar(cigar, inp_sequence, inp__quality)
self.assertEqual(exp_sequence, clipped_seq)
self.assertEqual(exp__quality, clipped_quality)
self.assertEqual({}, inserts)
def testSoftClip(self):
cigar = '3S6M'
inp_sequence = 'AAACAACCA'
inp__quality = 'BBBDDDEEE'
expect_sequence = 'CAACCA'
expect__quality = 'DDDEEE'
clipped_seq, clipped_quality, inserts = apply_cigar(cigar, inp_sequence, inp__quality)
self.assertEqual(expect_sequence, clipped_seq)
self.assertEqual(expect__quality, clipped_quality)
self.assertEqual({}, inserts)
def testInsertionLowQuality(self):
cigar = '3M3I6M'
inp_sequence = 'AAACAACCACCC'
inp__quality = 'BBBD*DEEEFFF'
exp_sequence = 'AAACCACCC'
exp__quality = 'BBBEEEFFF'
clipped_seq, clipped_quality, inserts = apply_cigar(cigar, inp_sequence, inp__quality)
self.assertEqual(exp_sequence, clipped_seq)
self.assertEqual(exp__quality, clipped_quality)
self.assertEqual({3: ('CAA', 'D*D')}, inserts)
def testLargeToken(self):
cigar = '12M'
inp_sequence = 'AAACAACCACCC'
inp__quality = 'BBBBBBBBBBBB'
exp_sequence = inp_sequence
exp__quality = inp__quality
clipped_seq, clipped_quality, inserts = apply_cigar(cigar, inp_sequence, inp__quality)
self.assertEqual(exp_sequence, clipped_seq)
self.assertEqual(exp__quality, clipped_quality)
self.assertEqual({}, inserts)
def testTrivial(self):
cigar = '9M'
inp_sequence = 'AAACAACCA'
inp__quality = 'BBBBBBBBB'
exp_sequence = inp_sequence
exp__quality = inp__quality
clipped_seq, clipped_quality, inserts = apply_cigar(cigar, inp_sequence, inp__quality)
self.assertEqual(exp_sequence, clipped_seq)
self.assertEqual(exp__quality, clipped_quality)
self.assertEqual({}, inserts)
def testPadding(self):
cigar = '12M'
seq = 'AAACAACCACCC' # @IgnorePep8
quality = 'BBBDDDEEEFFF' # @IgnorePep8
pos = 3
expected_seq = '---AAACAACCACCC' # @IgnorePep8
expected_quality = '!!!BBBDDDEEEFFF'
clipped_seq, clipped_quality, inserts = apply_cigar(cigar, seq, quality, pos)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testPadding(self):
cigar = '12M'
inp_sequence = 'AAACAACCACCC'
inp__quality = 'BBBDDDEEEFFF'
pos = 3
exp_sequence = '---AAACAACCACCC'
exp__quality = '!!!BBBDDDEEEFFF'
clipped_seq, clipped_quality, inserts = apply_cigar(cigar, inp_sequence, inp__quality, pos)
self.assertEqual(exp_sequence, clipped_seq)
self.assertEqual(exp__quality, clipped_quality)
self.assertEqual({}, inserts)
def testInsertionLowQuality(self):
cigar = '3M3I6M'
seq = 'AAACAACCACCC' # @IgnorePep8
quality = 'BBBD*DEEEFFF' # @IgnorePep8
expected_seq = 'AAACCACCC' # @IgnorePep8
expected_quality = 'BBBEEEFFF'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({3: ('CAA', 'D*D')}, inserts)
def testTrivial(self):
cigar = '9M'
seq = 'AAACAACCA' # @IgnorePep8
quality = 'BBBBBBBBB'
expected_seq = seq
expected_quality = quality
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testLargeToken(self):
cigar = '12M'
seq = 'AAACAACCACCC' # @IgnorePep8
quality = 'BBBBBBBBBBBB'
expected_seq = seq
expected_quality = quality
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testSoftClip(self):
cigar = '3S6M'
seq = 'AAACAACCA' # @IgnorePep8
quality = 'BBBDDDEEE' # @IgnorePep8
expected_seq = 'CAACCA' # @IgnorePep8
expected_quality = 'DDDEEE' # @IgnorePep8
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testDeletion(self):
cigar = '6M3D3M'
seq = 'AAACAACCA' # @IgnorePep8
quality = 'BBBDDDEEE' # @IgnorePep8
expected_seq = 'AAACAA---CCA' # @IgnorePep8
expected_quality = 'BBBDDD EEE'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testPadding(self):
cigar = '12M'
seq = 'AAACAACCACCC' # @IgnorePep8
quality = 'BBBDDDEEEFFF' # @IgnorePep8
pos = 3
expected_seq = '---AAACAACCACCC' # @IgnorePep8
expected_quality = '!!!BBBDDDEEEFFF'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality, pos)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testInsertionInsideClipRegionWithOffset(self):
cigar = '2M1I2M'
seq = 'TAGCT' # @IgnorePep8
quality = 'AABCC'
pos = 3
clip_from = 4
clip_to = 20
expected_seq = 'ACT' # @IgnorePep8
expected_quality = 'ACC'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality, pos, clip_from, clip_to)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({1: ('G', 'B')}, inserts)
def testInsertionBeforeClip(self):
cigar = '3M3I9M'
seq = 'AAAGGGCAACCACCC' # @IgnorePep8
quality = 'BBBHHHDDDEEEFFF' # @IgnorePep8
pos = 0
clip_from = 3
clip_to = 8
expected_seq = 'CAACCA' # @IgnorePep8
expected_quality = 'DDDEEE'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality, pos, clip_from, clip_to)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({0: ('GGG', 'HHH')}, inserts)
def testInsertionAfterClipping(self):
cigar = '3M3I3M'
seq = "ACTTAGAAA" # @IgnorePep8
quality = 'AAABBBDDD'
pos = 0
clip_from = 0
clip_to = 2
expected_seq = 'ACT' # @IgnorePep8
expected_quality = 'AAA'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality, pos, clip_from, clip_to)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testInsertionAtEndOfClipping(self):
cigar = '3M3I3M'
seq = "ACTTAGAAA" # @IgnorePep8
quality = 'AAABBBDDD'
pos = 0
clip_from = 0
clip_to = 3
expected_seq = 'ACTA' # @IgnorePep8
expected_quality = 'AAAD'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality, pos, clip_from, clip_to)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({3: ('TAG', 'BBB')}, inserts)
def testClipInsertionLowQuality(self):
cigar = '6M3I6M'
seq = 'AAACAAGGGCCACCC' # @IgnorePep8
quality = 'BBBDDDHH*EEEFFF' # @IgnorePep8
pos = 0
clip_from = 3
clip_to = 8
expected_seq = 'CAACCA' # @IgnorePep8
expected_quality = 'DDDEEE'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality, pos, clip_from, clip_to)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({3: ('GGG', 'HH*')}, inserts)
def testClippingEverything(self):
cigar = '12M'
seq = 'AAACAACCACCC' # @IgnorePep8
quality = 'BBBDDDEEEFFF' # @IgnorePep8
pos = 0
clip_from = 100
clip_to = 108
expected_seq = '' # @IgnorePep8
expected_quality = ''
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality, pos, clip_from, clip_to)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testInsertionAfterClipRegionWithOffset(self):
cigar = '5M1I2M'
seq = 'TAGCTCAG' # @IgnorePep8
quality = 'AAAAABCC'
pos = 10
clip_from = 10
clip_to = 13
expected_seq = 'TAGC' # @IgnorePep8
expected_quality = 'AAAA'
expected_inserts = {}
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar, seq, quality, pos, clip_from, clip_to)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual(expected_inserts, inserts)
def testInsertionAfterDeletion(self):
cigar = '3M3D3M3I3M'
(seq,
quality,
expected_seq,
expected_quality) = ('TTTCCCAAATTT',
'111222333444',
'TTT---CCCTTT',
'111 222444')
expected_inserts = {9: ('AAA', '333')}
seq, quality, inserts = apply_cigar(
cigar,
seq,
quality)
self.assertEqual(expected_seq, seq)
self.assertEqual(expected_quality, quality)
self.assertEqual(expected_inserts, inserts)
def testInsertionAfterInsertion(self):
cigar = '3M3I3M3I3M'
(seq,
quality,
expected_seq,
expected_quality) = ('TTTGGGCCCAAATTT',
'111222333444555',
'TTTCCCTTT',
'111333555')
expected_inserts = {3: ('GGG', '222'), 6: ('AAA', '444')}
seq, quality, inserts = apply_cigar(
cigar,
seq,
quality)
self.assertEqual(expected_seq, seq)
self.assertEqual(expected_quality, quality)
self.assertEqual(expected_inserts, inserts)
def testInsertionAtEndOfClipping(self):
cigar = '3M3I3M'
inp_sequence = "ACTTAGAAA"
inp__quality = 'AAABBBDDD'
pos = 0
clip_from = 0
clip_to = 3
exp_sequence = 'ACTA'
exp__quality = 'AAAD'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar,
inp_sequence,
inp__quality,
pos,
clip_from,
clip_to)
self.assertEqual(exp_sequence, clipped_seq)
self.assertEqual(exp__quality, clipped_quality)
self.assertEqual({3: ('TAG', 'BBB')}, inserts)
def testInsertionAfterClipping(self):
cigar = '3M3I3M'
inp_sequence = "ACTTAGAAA"
inp__quality = 'AAABBBDDD'
pos = 0
clip_from = 0
clip_to = 2
exp_sequence = 'ACT'
exp__quality = 'AAA'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar,
inp_sequence,
inp__quality,
pos,
clip_from,
clip_to)
self.assertEqual(exp_sequence, clipped_seq)
self.assertEqual(exp__quality, clipped_quality)
self.assertEqual({}, inserts)
def testInsertionAfterClipping(self):
cigar = '3M3I3M'
seq = "ACTTAGAAA" # @IgnorePep8
quality = 'AAABBBDDD'
pos = 0
clip_from = 0
clip_to = 2
expected_seq = 'ACT' # @IgnorePep8
expected_quality = 'AAA'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar,
seq,
quality,
pos,
clip_from,
clip_to)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({}, inserts)
def testInsertionAtEndOfClipping(self):
cigar = '3M3I3M'
seq = "ACTTAGAAA" # @IgnorePep8
quality = 'AAABBBDDD'
pos = 0
clip_from = 0
clip_to = 3
expected_seq = 'ACTA' # @IgnorePep8
expected_quality = 'AAAD'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar,
seq,
quality,
pos,
clip_from,
clip_to)
self.assertEqual(expected_seq, clipped_seq)
self.assertEqual(expected_quality, clipped_quality)
self.assertEqual({3: ('TAG', 'BBB')}, inserts)
def testClipInsertionLowQuality(self):
cigar = '6M3I6M'
inp_sequence = 'AAACAAGGGCCACCC'
inp__quality = 'BBBDDDHH*EEEFFF'
pos = 0
clip_from = 3
clip_to = 8
exp_sequence = 'CAACCA'
exp__quality = 'DDDEEE'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar,
inp_sequence,
inp__quality,
pos,
clip_from,
clip_to)
self.assertEqual(exp_sequence, clipped_seq)
self.assertEqual(exp__quality, clipped_quality)
self.assertEqual({3: ('GGG', 'HH*')}, inserts)
def testClippingEverything(self):
cigar = '12M'
inp_sequence = 'AAACAACCACCC'
inp__quality = 'BBBDDDEEEFFF'
pos = 0
clip_from = 100
clip_to = 108
exp_sequence = ''
exp__quality = ''
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar,
inp_sequence,
inp__quality,
pos,
clip_from,
clip_to)
self.assertEqual(exp_sequence, clipped_seq)
self.assertEqual(exp__quality, clipped_quality)
self.assertEqual({}, inserts)
def testInsertionInsideClipRegionWithOffset(self):
cigar = '2M1I2M'
inp_sequence = 'TAGCT'
inp__quality = 'AABCC'
pos = 3
clip_from = 4
clip_to = 20
exp_sequence = 'ACT'
exp__quality = 'ACC'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar,
inp_sequence,
inp__quality,
pos,
clip_from,
clip_to)
self.assertEqual(exp_sequence, clipped_seq)
self.assertEqual(exp__quality, clipped_quality)
self.assertEqual({1: ('G', 'B')}, inserts)
def testInsertionBeforeClip(self):
cigar = '3M3I9M'
inp_sequence = 'AAAGGGCAACCACCC'
inp__quality = 'BBBHHHDDDEEEFFF'
pos = 0
clip_from = 3
clip_to = 8
exp_sequence = 'CAACCA'
exp__quality = 'DDDEEE'
clipped_seq, clipped_quality, inserts = apply_cigar(
cigar,
inp_sequence,
inp__quality,
pos,
clip_from,
clip_to)
self.assertEqual(exp_sequence, clipped_seq)
self.assertEqual(exp__quality, clipped_quality)
self.assertEqual({0: ('GGG', 'HHH')}, inserts)