def main():
parser = argparse.ArgumentParser()
parser.add_argument('input',
help="Bam file in order of query name - for stdin")
args = parser.parse_args()
inf = sys.stdin
if args.input != '-':
cmd = "samtools view -h " + args.input
p = Popen(cmd.split(), stdout=PIPE)
inf = p.stdout
mesr = MultiEntrySamReader(inf)
while True:
entries = mesr.read_entries()
if not entries: break
if len(entries) == 0: break
if entries[0].value('cigar') == '*':
print entries[0].value('qname') + "\t0"
continue
sam = entries[0]
m = re.search('NH:i:(\d+)', sam.entry['remainder'])
if not m:
sys.stderr.write("ERROR not a hisat entry\n")
sys.exit()
cnt = max([len(entries), int(m.group(1))])
print entries[0].value('qname') + "\t" + str(cnt)
def main():
parser = argparse.ArgumentParser(
description=
"Take a sam file and join together mate pairs into single alignments. Alignments must be ordered by query name."
)
parser.add_argument(
'input', help="FILENAME input .sam or .bam or '-' for STDIN sam")
group = parser.add_mutually_exclusive_group()
group.add_argument('--sam', action='store_true')
group.add_argument('--bam', action='store_true')
parser.add_argument('--mates_only',
action='store_true',
help="Only output combined mates")
parser.add_argument('--threads',
type=int,
default=1,
help="Number of threads to use, default is 1")
args = parser.parse_args()
inf = sys.stdin
if args.bam or (not args.sam and not args.input == '-'):
fh = open(args.input)
p = Popen('samtools view - -h'.split(), stdin=fh, stdout=PIPE)
inf = p.stdout
buffer_size = 10000
buffer = []
msr = MultiEntrySamReader(inf)
spc = SAMtoPSLconversionFactory()
psc = PSLtoSAMconversionFactory()
# set the headers for the spc
for h in msr.header:
print h.rstrip()
spc.read_header_line(h)
if args.threads > 1:
p1 = Pool(processes=args.threads)
while True:
entries = msr.read_entries()
if not entries: break
buffer.append(entries)
if len(buffer) >= buffer_size:
if args.threads > 1:
p1.apply_async(do_buffer,
args=(buffer, msr, spc, psc, args),
callback=do_callback)
else:
v = do_buffer(buffer, msr, spc, psc, args)
do_callback(v)
buffer = []
if len(buffer) > 0:
if args.threads > 1:
p1.apply_async(do_buffer,
args=(buffer, msr, spc, psc, args),
callback=do_callback)
else:
v = do_buffer(buffer, msr, spc, psc, args)
do_callback(v)
if args.threads > 1:
p1.close()
p1.join()
def main():
parser = argparse.ArgumentParser(
description="Take a sam or bam file and output the best alignment for each read, it still can output the same read name twice if they happen to be mate pairs, but it will only output the best alignment for each individual mate, not necessarily the two together. You could combine mates if that is helpful with another script."
)
parser.add_argument("input", help="FILENAME input .sam or .bam or '-' for STDIN sam")
group = parser.add_mutually_exclusive_group()
group.add_argument("--bam", action="store_true")
group.add_argument("--sam", action="store_true")
args = parser.parse_args()
inf = sys.stdin
if args.bam or (not args.sam and not args.input == "-"):
fh = open(args.input)
p = Popen("samtools view - -h".split(), stdin=fh, stdout=PIPE)
inf = p.stdout
msr = MultiEntrySamReader(inf)
spc = SAMtoPSLconversionFactory()
# set the headers for the spc
for h in msr.header:
print h.rstrip()
spc.read_header_line(h)
while True:
entries = msr.read_entries()
if not entries:
break
longest0 = 0
entry0 = None
longest1 = 0
entry1 = None
longest2 = 0
entry2 = None
for sam in entries:
pline = spc.convert_line(sam.get_line())
if not pline:
continue
side = None
if sam.check_flag(64):
side = 1
if sam.check_flag(128):
side = 2
p = PSL(pline)
if p.get_coverage() > longest0:
longest0 = p.get_coverage()
entry0 = sam
if side == 1 and p.get_coverage() > longest1:
longest1 = p.get_coverage()
entry1 = sam
if side == 2 and p.get_coverage() > longest2:
longest2 = p.get_coverage()
entry2 = sam
if entry0: # output the combined if its there
print entry0.get_line()
else:
if entry1: # output each of the mates if they are paired but not joined
print entry1.get_line()
if entry2:
print entry2.get_line()
def main():
parser = argparse.ArgumentParser(
description=
"Take a sam or bam file and output the best alignment for each read, it still can output the same read name twice if they happen to be mate pairs, but it will only output the best alignment for each individual mate, not necessarily the two together. You could combine mates if that is helpful with another script."
)
parser.add_argument(
'input', help="FILENAME input .sam or .bam or '-' for STDIN sam")
group = parser.add_mutually_exclusive_group()
group.add_argument('--bam', action='store_true')
group.add_argument('--sam', action='store_true')
args = parser.parse_args()
inf = sys.stdin
if args.bam or (not args.sam and not args.input == '-'):
fh = open(args.input)
p = Popen('samtools view - -h'.split(), stdin=fh, stdout=PIPE)
inf = p.stdout
msr = MultiEntrySamReader(inf)
spc = SAMtoPSLconversionFactory()
# set the headers for the spc
for h in msr.header:
print h.rstrip()
spc.read_header_line(h)
while True:
entries = msr.read_entries()
if not entries: break
longest0 = 0
entry0 = None
longest1 = 0
entry1 = None
longest2 = 0
entry2 = None
for sam in entries:
pline = spc.convert_line(sam.get_line())
if not pline: continue
side = None
if sam.check_flag(64): side = 1
if sam.check_flag(128): side = 2
p = PSL(pline)
if p.get_coverage() > longest0:
longest0 = p.get_coverage()
entry0 = sam
if side == 1 and p.get_coverage() > longest1:
longest1 = p.get_coverage()
entry1 = sam
if side == 2 and p.get_coverage() > longest2:
longest2 = p.get_coverage()
entry2 = sam
if entry0: #output the combined if its there
print entry0.get_line()
else:
if entry1: #output each of the mates if they are paired but not joined
print entry1.get_line()
if entry2:
print entry2.get_line()
def main():
parser = argparse.ArgumentParser(description="Take a sam file and join together mate pairs into single alignments. Alignments must be ordered by query name.")
parser.add_argument('input',help="FILENAME input .sam or .bam or '-' for STDIN sam")
group = parser.add_mutually_exclusive_group()
group.add_argument('--sam',action='store_true')
group.add_argument('--bam',action='store_true')
parser.add_argument('--mates_only',action='store_true',help="Only output combined mates")
parser.add_argument('--threads',type=int,default=1,help="Number of threads to use, default is 1")
args = parser.parse_args()
inf = sys.stdin
if args.bam or (not args.sam and not args.input == '-'):
fh = open(args.input)
p = Popen('samtools view - -h'.split(),stdin=fh,stdout=PIPE)
inf = p.stdout
buffer_size = 10000
buffer = []
msr = MultiEntrySamReader(inf)
spc = SAMtoPSLconversionFactory()
psc = PSLtoSAMconversionFactory()
# set the headers for the spc
for h in msr.header:
print h.rstrip()
spc.read_header_line(h)
if args.threads > 1:
p1 = Pool(processes=args.threads)
while True:
entries = msr.read_entries()
if not entries: break
buffer.append(entries)
if len(buffer) >= buffer_size:
if args.threads > 1:
p1.apply_async(do_buffer,args=(buffer,msr,spc,psc,args),callback=do_callback)
else:
v = do_buffer(buffer,msr,spc,psc,args)
do_callback(v)
buffer = []
if len(buffer) > 0:
if args.threads > 1:
p1.apply_async(do_buffer,args=(buffer,msr,spc,psc,args),callback=do_callback)
else:
v = do_buffer(buffer,msr,spc,psc,args)
do_callback(v)
if args.threads > 1:
p1.close()
p1.join()
def main():
parser = argparse.ArgumentParser()
parser.add_argument('input',help="Bam file in order of query name - for stdin")
args = parser.parse_args()
inf = sys.stdin
if args.input != '-':
cmd = "samtools view -h "+args.input
p = Popen(cmd.split(),stdout=PIPE)
inf = p.stdout
mesr = MultiEntrySamReader(inf)
while True:
entries = mesr.read_entries()
if not entries: break
if len(entries) == 0: break
if entries[0].value('cigar') == '*':
print entries[0].value('qname')+"\t0"
continue
sam = entries[0]
m = re.search('NH:i:(\d+)',sam.entry['remainder'])
if not m:
sys.stderr.write("ERROR not a hisat entry\n")
sys.exit()
cnt = max([len(entries),int(m.group(1))])
print entries[0].value('qname')+"\t"+str(cnt)
