def main():
parser = argparse.ArgumentParser(
description="", formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('input', help="Use BAM or - for SAM STDIN")
parser.add_argument('-o', '--output', help="output file")
args = parser.parse_args()
of = sys.stdout
if args.output:
if re.search('\.gz$', args.output):
of = gzip.open(args.output, 'w')
else:
of = open(args.output, 'w')
inf = sys.stdin
if args.input != '-':
cmd = "samtools view " + args.input
p = Popen(cmd.split(), stdout=PIPE)
inf = p.stdout
for line in inf:
v = SAM(line)
for rng in [x.get_range() for x in v.get_target_transcript(68).exons]:
of.write("\t".join([str(x) for x in rng.get_bed_array()]) + "\n")
if args.input != '-':
p.communicate()
of.close()
def main():
parser = argparse.ArgumentParser(description="",formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('input',help="Use BAM or - for SAM STDIN")
parser.add_argument('-o','--output',help="output file")
args = parser.parse_args()
of = sys.stdout
if args.output:
if re.search('\.gz$',args.output):
of = gzip.open(args.output,'w')
else:
of = open(args.output,'w')
inf = sys.stdin
if args.input != '-':
cmd = "samtools view "+args.input
p = Popen(cmd.split(),stdout=PIPE)
inf = p.stdout
for line in inf:
v = SAM(line)
for rng in [x.get_range() for x in v.get_target_transcript(68).exons]:
of.write("\t".join([str(x) for x in rng.get_bed_array()])+"\n")
if args.input != '-':
p.communicate()
of.close()
def get_SAM(self, min_intron_size=68):
from Bio.Format.Sam import SAM
#ar is target then query
qname = self.get_alignment_ranges()[0][1].chr
flag = 0
if self.get_strand() == '-': flag = 16
rname = self.get_alignment_ranges()[0][0].chr
pos = self.get_alignment_ranges()[0][0].start
mapq = 255
cigar = self.construct_cigar(min_intron_size)
rnext = '*'
pnext = 0
tlen = self.get_target_range().length()
seq = self.get_query_sequence()
if not seq: seq = '*'
qual = self.get_query_quality()
if not qual: qual = '*'
#seq = '*'
#qual = '*'
if self.get_strand() == '-':
seq = rc(seq)
qual = qual[::-1]
ln = qname + "\t" + str(flag) + "\t" + rname + "\t" + \
str(pos) + "\t" + str(mapq) + "\t" + cigar + "\t" + \
rnext + "\t" + str(pnext) + "\t" + str(tlen) + "\t" + \
seq + "\t" + qual
return SAM(ln, reference=self._reference)
def do_seq(seq,args,fname):
cmd = 'samtools view '+args.input+' '+seq
p = Popen(cmd.split(),stdout=PIPE)
cmd = 'bedtools genomecov -i - -bg -g '+fname
po2 = Popen(cmd.split(),stdin=PIPE,stdout=PIPE)
cmd = 'sort -k 1,1 -k2,2n -k3,3n -S1G --parallel='+str(args.threads)
if args.tempdir: cmd += ' -T '+args.tempdir
po1 = Popen(cmd.split(),stdin=PIPE,stdout=po2.stdin)
for line in p.stdout:
sam = SAM(line)
if not sam.is_aligned(): continue
for ex in sam.get_target_transcript(min_intron=68).exons:
bed = "\t".join([str(x) for x in ex.get_range().get_bed_array()])
po1.stdin.write(bed+"\n")
po1.communicate()
res = po2.communicate()[0]
return res
def do_seq(seq, args, fname):
cmd = 'samtools view ' + args.input + ' ' + seq
p = Popen(cmd.split(), stdout=PIPE)
cmd = 'bedtools genomecov -i - -bg -g ' + fname
po2 = Popen(cmd.split(), stdin=PIPE, stdout=PIPE)
cmd = 'sort -k 1,1 -k2,2n -k3,3n -S1G --parallel=' + str(args.threads)
if args.tempdir: cmd += ' -T ' + args.tempdir
po1 = Popen(cmd.split(), stdin=PIPE, stdout=po2.stdin)
for line in p.stdout:
sam = SAM(line)
if not sam.is_aligned(): continue
for ex in sam.get_target_transcript(min_intron=68).exons:
bed = "\t".join([str(x) for x in ex.get_range().get_bed_array()])
po1.stdin.write(bed + "\n")
po1.communicate()
res = po2.communicate()[0]
return res
