def clip_primer_overlap_bases(alignedread, primer_overlap_peak_pos, primer_len):
"""Clip alignread's bases overlapping with primer pos by changing
cigar string
"""
if alignedread.is_reverse:
# WARNING: partial code duplication with else branch
#
# this is a reverse read overlapping with it's 5p end,
# so starting from there clip bases, skipping any
# non-match ops. do this p+primer_len -
# alignedread.pos + 1 times
num_clip = primer_overlap_peak_pos + primer_len - alignedread.pos + 1
LOG.debug("rv read starting at %d overlapping with fw peak start %d+%d by %d: %s" % (
alignedread.pos, primer_overlap_peak_pos, primer_len, num_clip, alignedread))
new_cigar_decoded = []
for op in cigar.decoded_ops(alignedread.cigar):
if num_clip == 0 or op == 'D':
new_cigar_decoded.append(op)
else:
# replace everything else with S
new_cigar_decoded.append('S')
if op in 'MI=X':
num_clip -= 1
else:
# WARNING: partial code duplication with if branch
#
# this is a forward read overlapping with its' 3p end, so
# starting from there clip bases, skipping any non-match
# ops. do this p+primer_len - alignedread.pos + 1 times
num_clip = alignedread.aend - primer_overlap_peak_pos + primer_len + 1
LOG.debug("fw read ending at %d overlapping with rv peak end %d-%d by %d: %s" % (
alignedread.aend, primer_overlap_peak_pos, primer_len, num_clip, alignedread))
new_cigar_decoded = []
for op in list(cigar.decoded_ops(alignedread.cigar))[::-1]:
if num_clip == 0 or op == 'D':
new_cigar_decoded.insert(0, op)
else:
# replace everything else with S
new_cigar_decoded.insert(0, 'S')
if op in 'MI=X':
num_clip -= 1
new_cigar = cigar.parse(
cigar.cigar_from_decoded_ops(new_cigar_decoded))
# According to spec http://samtools.sourceforge.net/SAM1.pdf:
# "Sum of lengths of the M/I/S/=/X operations shall equal the
# length of SEQ"
cigar_len = sum([1 for op in new_cigar_decoded if op in 'MISX='])
assert cigar_len == alignedread.rlen, (
"read length derived from new cigar (%s -> %d) mismatches rlen=%d for %s read: %s" % (
new_cigar, cigar_len, alignedread.rlen,
"reverse" if alignedread.is_reverse else "forward", alignedread))
alignedread.cigar = new_cigar
def print_alignedread_info(alignedread):
"""Debugging only"""
print "- pos, aend-1, alen = %d %d %s" % (
alignedread.pos, alignedread.aend-1, alignedread.alen)
print "- cigar = %s" % (alignedread.cigar)
print "- decoded cigar = %s" % (
''.join(list(cigar.decoded_ops(alignedread.cigar))))
try:
alnlen = cigar.aligned_length(alignedread.cigar)
except ValueError:
alnlen = -1
print "- cigar aln-len = %d" % (alnlen)
print "- read sequence = %s" % (alignedread.query)
print "- aligned parts = %s" % (alignedread.query)