def main():
parser = argparse.ArgumentParser(
description="", formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('input', help="BAM file or Use - for STDIN for SAM")
parser.add_argument('--minimum_intron',
type=int,
default=68,
help="smallest intron")
parser.add_argument('-o', '--output', help="Output file, gzip is okay")
args = parser.parse_args()
of = sys.stdout
if args.output:
if args.output[-3:] == '.gz':
of = gzip.open(args.output, 'w')
else:
of = open(args.output, 'w')
if args.input == '-':
sh = SamStream(sys.stdin)
else:
sh = BAMFile(args.input)
for e in sh:
if not e.is_aligned(): continue
gpd_line = e.get_target_transcript(
min_intron=args.minimum_intron).get_gpd_line()
of.write(gpd_line + "\n")
sh.close()
of.close()
def main():
parser = argparse.ArgumentParser(description="",formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('input',help="Use bam file")
parser.add_argument('output',help="Use bam file")
parser.add_argument('--threads',type=int,default=cpu_count(),help="Thread count")
args = parser.parse_args()
m = re.match('^(\S+)\.bam$',args.output)
if not m:
sys.stderr.write("use bam output")
sys.exit()
cmd1 = 'samtools view -H '+args.input
p1 = Popen(cmd1.split(),stdout=PIPE)
bs = SamStream(p1.stdout)
rlens = bs.get_header().get_sequence_lengths()
htext = bs.header_text
p1.communicate()
hlines = htext.rstrip().split("\n")
done_lens = False
cmd = 'samtools sort -@ '+str(args.threads)+' - '+m.group(1)
sys.stderr.write(cmd+"\n")
p = Popen(cmd.split(),stdin=PIPE)
for ln in hlines:
if re.match('@SQ\tSN:',ln):
if not done_lens:
done_lens = True
for chr in sorted(rlens.keys()):
p.stdin.write("@SQ\tSN:"+chr+"\tLN:"+str(rlens[chr])+"\n")
else:
p.stdin.write(ln.rstrip("\n")+"\n")
cmd1 = 'samtools view '+args.input
p1 = Popen(cmd1.split(),stdout=p.stdin)
p1.communicate()
p.communicate()
def main():
parser = argparse.ArgumentParser(
description="", formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('input', help="Use bam file")
parser.add_argument('output', help="Use bam file")
parser.add_argument('--threads',
type=int,
default=cpu_count(),
help="Thread count")
args = parser.parse_args()
m = re.match('^(\S+)\.bam$', args.output)
if not m:
sys.stderr.write("use bam output")
sys.exit()
cmd1 = 'samtools view -H ' + args.input
p1 = Popen(cmd1.split(), stdout=PIPE)
bs = SamStream(p1.stdout)
rlens = bs.get_header().get_sequence_lengths()
htext = bs.header_text
p1.communicate()
hlines = htext.rstrip().split("\n")
done_lens = False
cmd = 'samtools sort -@ ' + str(args.threads) + ' - ' + m.group(1)
sys.stderr.write(cmd + "\n")
p = Popen(cmd.split(), stdin=PIPE)
for ln in hlines:
if re.match('@SQ\tSN:', ln):
if not done_lens:
done_lens = True
for chr in sorted(rlens.keys()):
p.stdin.write("@SQ\tSN:" + chr + "\tLN:" +
str(rlens[chr]) + "\n")
else:
p.stdin.write(ln.rstrip("\n") + "\n")
cmd1 = 'samtools view ' + args.input
p1 = Popen(cmd1.split(), stdout=p.stdin)
p1.communicate()
p.communicate()
def main():
parser = argparse.ArgumentParser(description="",formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('input',help="BAM file or Use - for STDIN for SAM")
parser.add_argument('--minimum_intron',type=int,default=68,help="smallest intron")
parser.add_argument('-o','--output',help="Output file, gzip is okay")
args = parser.parse_args()
of = sys.stdout
if args.output:
if args.output[-3:]=='.gz':
of = gzip.open(args.output,'w')
else:
of = open(args.output,'w')
if args.input =='-':
sh = SamStream(sys.stdin)
else:
sh = BAMFile(args.input)
for e in sh:
if not e.is_aligned(): continue
gpd_line = e.get_target_transcript(min_intron=args.minimum_intron).get_gpd_line()
of.write(gpd_line+"\n")
sh.close()
of.close()
def do_sam(args):
if args.input != '-':
m = re.search('\.bam$',args.input)
if not m:
sys.stderr.write("ERROR input expects bam unless piping to stdin.. then SAM with header\n")
sys.exit()
if not args.output:
sys.stderr.write("ERROR sam sorts must output to a bam file\n")
sys.exit()
m = re.match('^(.+)\.bam$',args.output)
if not m:
sys.stderr.write("ERROR sam sorts must output to a bam file\n")
sys.exit()
cmdout = 'samtools sort - '+m.group(1)
if args.threads: cmdout += ' -@ '+str(args.threads)
inf = None
if args.input == '-':
inf = sys.stdin
else:
cmd = 'samtools view -h '+args.input
p = Popen(cmd.split(),stdout=PIPE,bufsize=1)
inf = p.stdout
s = SamStream(inf)
header = s.header_text.rstrip().split("\n")
split_stream = [header[i].split("\t") for i in range(0,len(header))]
sq_inds = [i for i in range(0,len(split_stream)) if split_stream[i][0]=='@SQ']
nonsq_inds = [i for i in range(0,len(split_stream)) if split_stream[i][0]!='@SQ']
top = [header[i] for i in nonsq_inds]
chroms = sorted([split_stream[i] for i in sq_inds],key = lambda x: x[1][3:])
cmd2 = 'samtools view -Sb -'
pout = Popen(cmdout.split(),stdin=PIPE)
p2 = Popen(cmd2.split(),stdin=PIPE,stdout=pout.stdin)
for t in top:
p2.stdin.write(t.rstrip()+"\n")
for c in chroms:
p2.stdin.write("\t".join(c).rstrip()+"\n")
for sam in s:
p2.stdin.write(sam.get_line().rstrip()+"\n")
p2.communicate()
pout.communicate()
if args.input != '-':
p.communicate()
return
def main():
parser = argparse.ArgumentParser(description="",formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('input',help="Use - for STDIN or read bam")
parser.add_argument('--minimum_intron_size',type=int,default=68,help="Require intron to be this size or larger")
parser.add_argument('--minimum_overhang',type=int,default=10,help="At least this many bases on each side of an intron")
parser.add_argument('--minimum_support',type=int,default=1,help="Minimum number of reads that should support a junction to report any of the reads")
args = parser.parse_args()
inf = sys.stdin
if args.input != '-':
cmd = 'samtools view -F 4 -h '+args.input
p = Popen(cmd.split(),stdout=PIPE,bufsize=1)
inf = p.stdout
cmd2 = 'awk '+"'"+'{if(NF<10) print $0; else if($6~/N/) print $0;}'+"'"
p2 = Popen(cmd2,stdout=PIPE,stdin=inf,bufsize=1,shell=True)
stream = SamStream(p2.stdout,minimum_intron_size=args.minimum_intron_size,minimum_overhang=args.minimum_overhang)
lstream = LocusStream(stream)
for h in stream.header:
print h.rstrip()
for r in lstream:
# now we have all the possible junctions from the range
[juncs,sams] = get_junctions(r.get_payload(),args)
evidence = {}
lines = {}
for x in set([x[0].get_range_string() for x in juncs]):
evidence[x] = 0
lines[x] = set()
for i in range(0,len(juncs)):
jstr = juncs[i][0].get_range_string()
evidence[jstr]+=1
lines[jstr].add(juncs[i][1])
accepted = set()
for jstr in evidence:
if evidence[jstr] >= args.minimum_support:
#print jstr
for i in lines[jstr]: accepted.add(i)
for i in sorted(list(accepted)):
print sams[i].get_line().rstrip()
p2.communicate()
if args.input != '-':
p.communicate()
def main():
parser = argparse.ArgumentParser(
description="", formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('input', help="Use - for STDIN or specify a BAM file")
parser.add_argument('-r',
'--reference',
help="Reference fasta",
required=True)
args = parser.parse_args()
ref = None
if args.reference:
ref = FastaData(open(args.reference, 'rb').read())
if args.input == '-':
args.input = SamStream(sys.stdin, reference=ref)
else:
args.input = BAMFile(args.input, reference=ref)
for e in args.input:
if e.is_aligned():
print e.get_PSL()