def genrefblks(readseq, chrom, start, stop, strand, cigar, nreads):
refpos = start
readpos = 0
if strand == '-':
readseq = reverse_complement(readseq)
tleftlim, trightlim = start + ENDTRIM, stop - ENDTRIM
qleftlim, qrightlim = ENDTRIM, len(readseq) - ENDTRIM
cigarcommands = cigar_pattern.findall(cigar)
if cigarcommands[0][1] == 'S': # shift start site for the first soft clipping
start -= int(cigarcommands[0][0])
if len(cigarcommands) > 1 and cigarcommands[-1][1] == 'S': # last soft clipping
stop += int(cigarcommands[-1][0])
for num, cmd in cigarcommands:
num = int(num)
if cmd == 'M': # match
mleft = max(qleftlim, readpos)
mright = min(qrightlim, readpos + num)
if mleft < mright:
seq = readseq[mleft:mright]
yield ('M', nreads, max(refpos, tleftlim), seq)
refpos += num
readpos += num
elif cmd == 'S': # soft clip
readpos += num
elif cmd == 'N': # skip
refpos += num
elif cmd == 'D': # deletion
if tleftlim <= refpos < trightlim:
yield ('D', nreads, refpos, num)
refpos += num
elif cmd == 'I': # insertion
ppos = (refpos if strand == '+' else (refpos-1))
if tleftlim <= ppos < trightlim:
yield ('I', nreads, ppos, num)
readpos += num
elif cmd == 'H': # hard clipping
pass
else:
yield ('E', nreads, num, cmd, readseq)
raise ValueError
if strand == '+':
fivep, threep = start, stop-1
else:
fivep, threep = stop-1, start
yield ('5', nreads, fivep)
yield ('3', nreads, threep)
def get_refseq(self, chrom, start, stop): # with automatic (-) strand detection
if start >= 0:
return self.seqs.get(chrom, start, stop).upper()
start, stop = -stop, -start
return sequtils.reverse_complement(self.seqs.get(chrom, start, stop).upper())