def test_check_is_left():
assert check_is_left("seq1/1")
assert check_is_left("seq1 1::N")
assert not check_is_left("seq1/2")
assert not check_is_left("seq1 2::N")
assert not check_is_left("seq")
assert not check_is_left("seq 1")
assert check_is_left("@HWI-ST412:261:d15khacxx:8:1101:3149:2157 1:N:0:ATCACG")
def test_check_is_left():
assert check_is_left('seq1/1')
assert check_is_left('seq1 1::N')
assert not check_is_left('seq1/2')
assert not check_is_left('seq1 2::N')
assert not check_is_left('seq')
assert not check_is_left('seq 1')
assert check_is_left(
'@HWI-ST412:261:d15khacxx:8:1101:3149:2157 1:N:0:ATCACG')
def test_check_is_left():
assert check_is_left('seq1/1')
assert check_is_left('seq1 1::N')
assert not check_is_left('seq1/2')
assert not check_is_left('seq1 2::N')
assert not check_is_left('seq')
assert not check_is_left('seq 1')
assert check_is_left(
'@HWI-ST412:261:d15khacxx:8:1101:3149:2157 1:N:0:ATCACG')
def main():
info('interleave-reads.py')
args = sanitize_help(get_parser()).parse_args()
check_input_files(args.left, args.force)
check_input_files(args.right, args.force)
check_space([args.left, args.right], args.force)
s1_file = args.left
s2_file = args.right
print("Interleaving:\n\t%s\n\t%s" % (s1_file, s2_file), file=sys.stderr)
outfp = get_file_writer(args.output, args.gzip, args.bzip)
counter = 0
screed_iter_1 = screed.open(s1_file)
screed_iter_2 = screed.open(s2_file)
for read1, read2 in zip_longest(screed_iter_1, screed_iter_2):
if read1 is None or read2 is None:
print(("ERROR: Input files contain different number"
" of records."),
file=sys.stderr)
sys.exit(1)
if counter % 100000 == 0:
print('...', counter, 'pairs', file=sys.stderr)
counter += 1
name1 = read1.name
name2 = read2.name
if not args.no_reformat:
if not check_is_left(name1):
name1 += '/1'
if not check_is_right(name2):
name2 += '/2'
read1.name = name1
read2.name = name2
if not check_is_pair(read1, read2):
print("ERROR: This doesn't look like paired data! "
"%s %s" % (read1.name, read2.name),
file=sys.stderr)
sys.exit(1)
write_record_pair(read1, read2, outfp)
print('final: interleaved %d pairs' % counter, file=sys.stderr)
print('output written to', describe_file_handle(outfp), file=sys.stderr)
def main():
info('interleave-reads.py')
args = sanitize_help(get_parser()).parse_args()
check_input_files(args.left, args.force)
check_input_files(args.right, args.force)
check_space([args.left, args.right], args.force)
s1_file = args.left
s2_file = args.right
fail = False
print("Interleaving:\n\t%s\n\t%s" % (s1_file, s2_file), file=sys.stderr)
outfp = get_file_writer(args.output, args.gzip, args.bzip)
counter = 0
screed_iter_1 = screed.open(s1_file)
screed_iter_2 = screed.open(s2_file)
for read1, read2 in zip_longest(screed_iter_1, screed_iter_2):
if read1 is None or read2 is None:
print(("ERROR: Input files contain different number"
" of records."), file=sys.stderr)
sys.exit(1)
if counter % 100000 == 0:
print('...', counter, 'pairs', file=sys.stderr)
counter += 1
name1 = read1.name
name2 = read2.name
if not args.no_reformat:
if not check_is_left(name1):
name1 += '/1'
if not check_is_right(name2):
name2 += '/2'
read1.name = name1
read2.name = name2
if not check_is_pair(read1, read2):
print("ERROR: This doesn't look like paired data! "
"%s %s" % (read1.name, read2.name), file=sys.stderr)
sys.exit(1)
write_record_pair(read1, read2, outfp)
print('final: interleaved %d pairs' % counter, file=sys.stderr)
print('output written to', describe_file_handle(outfp), file=sys.stderr)
def main():
info('interleave-reads.py')
args = get_parser().parse_args()
for _ in args.infiles:
check_file_status(_, args.force)
check_space(args.infiles, args.force)
s1_file = args.infiles[0]
if len(args.infiles) == 2:
s2_file = args.infiles[1]
else:
s2_file = s1_file.replace('_R1_', '_R2_')
if s1_file == s2_file:
print >>sys.stderr, ("ERROR: given only one filename, that "
"doesn't contain _R1_. Exiting.")
sys.exit(1)
print >> sys.stderr, ("given only one file; "
"guessing that R2 file is %s" % s2_file)
fail = False
if not os.path.exists(s1_file):
print >> sys.stderr, "Error! R1 file %s does not exist" % s1_file
fail = True
if not os.path.exists(s2_file):
print >> sys.stderr, "Error! R2 file %s does not exist" % s2_file
fail = True
if fail and not args.force:
sys.exit(1)
print >> sys.stderr, "Interleaving:\n\t%s\n\t%s" % (s1_file, s2_file)
counter = 0
screed_iter_1 = screed.open(s1_file, parse_description=False)
screed_iter_2 = screed.open(s2_file, parse_description=False)
for read1, read2 in itertools.izip_longest(screed_iter_1, screed_iter_2):
if read1 is None or read2 is None:
print >>sys.stderr, ("ERROR: Input files contain different number"
" of records.")
sys.exit(1)
if counter % 100000 == 0:
print >> sys.stderr, '...', counter, 'pairs'
counter += 1
name1 = read1.name
if not check_is_left(name1):
name1 += '/1'
name2 = read2.name
if not check_is_right(name2):
name2 += '/2'
read1.name = name1
read2.name = name2
if not check_is_pair(read1, read2):
print >>sys.stderr, "ERROR: This doesn't look like paired data! " \
"%s %s" % (read1.name, read2.name)
sys.exit(1)
write_record_pair(read1, read2, args.output)
print >> sys.stderr, 'final: interleaved %d pairs' % counter
print >> sys.stderr, 'output written to', args.output.name
def main():
info('split-paired-reads.py')
args = get_parser().parse_args()
infile = args.infile
check_input_files(infile, args.force)
filenames = [infile]
check_space(filenames, args.force)
# decide where to put output files - specific directory? or just default?
if args.output_directory:
if not os.path.exists(args.output_directory):
os.makedirs(args.output_directory)
out1 = args.output_directory + '/' + os.path.basename(infile) + '.1'
out2 = args.output_directory + '/' + os.path.basename(infile) + '.2'
else:
out1 = os.path.basename(infile) + '.1'
out2 = os.path.basename(infile) + '.2'
# OVERRIDE output file locations with -1, -2
if args.output_first:
out1 = args.output_first
if args.output_second:
out2 = args.output_second
fp_out1 = open(out1, 'w')
fp_out2 = open(out2, 'w')
counter1 = 0
counter2 = 0
index = None
screed_iter = screed.open(infile, parse_description=False)
# walk through all the reads in broken-paired mode.
for index, is_pair, record1, record2 in broken_paired_reader(screed_iter):
if index % 100000 == 0 and index:
print >> sys.stderr, '...', index
# are we requiring pairs?
if args.force_paired and not is_pair:
print >>sys.stderr, 'ERROR, %s is not part of a pair' % \
record1.name
sys.exit(1)
if is_pair:
write_record(record1, fp_out1)
counter1 += 1
write_record(record2, fp_out2)
counter2 += 1
else:
name = record1.name
if check_is_left(name):
write_record(record1, fp_out1)
counter1 += 1
elif check_is_right(name):
write_record(record1, fp_out2)
counter2 += 1
else:
print >>sys.stderr, \
"Unrecognized format for read pair information: %s" % name
print >> sys.stderr, "Exiting."
sys.exit(1)
print >> sys.stderr, "DONE; split %d sequences (%d left, %d right)" % \
(counter1 + counter2, counter1, counter2)
print >> sys.stderr, "/1 reads in %s" % out1
print >> sys.stderr, "/2 reads in %s" % out2
def main():
info('split-paired-reads.py')
args = get_parser().parse_args()
infile = args.infile
check_input_files(infile, args.force)
filenames = [infile]
check_space(filenames, args.force)
# decide where to put output files - specific directory? or just default?
if args.output_directory:
if not os.path.exists(args.output_directory):
os.makedirs(args.output_directory)
out1 = args.output_directory + '/' + os.path.basename(infile) + '.1'
out2 = args.output_directory + '/' + os.path.basename(infile) + '.2'
else:
out1 = os.path.basename(infile) + '.1'
out2 = os.path.basename(infile) + '.2'
# OVERRIDE output file locations with -1, -2
if args.output_first:
out1 = args.output_first
if args.output_second:
out2 = args.output_second
fp_out1 = open(out1, 'w')
fp_out2 = open(out2, 'w')
counter1 = 0
counter2 = 0
index = None
screed_iter = screed.open(infile, parse_description=False)
# walk through all the reads in broken-paired mode.
for index, is_pair, record1, record2 in broken_paired_reader(screed_iter):
if index % 100000 == 0 and index:
print >> sys.stderr, '...', index
# are we requiring pairs?
if args.force_paired and not is_pair:
print >>sys.stderr, 'ERROR, %s is not part of a pair' % \
record1.name
sys.exit(1)
if is_pair:
write_record(record1, fp_out1)
counter1 += 1
write_record(record2, fp_out2)
counter2 += 1
else:
name = record1.name
if check_is_left(name):
write_record(record1, fp_out1)
counter1 += 1
elif check_is_right(name):
write_record(record1, fp_out2)
counter2 += 1
else:
print >>sys.stderr, \
"Unrecognized format for read pair information: %s" % name
print >>sys.stderr, "Exiting."
sys.exit(1)
print >> sys.stderr, "DONE; split %d sequences (%d left, %d right)" % \
(counter1 + counter2, counter1, counter2)
print >> sys.stderr, "/1 reads in %s" % out1
print >> sys.stderr, "/2 reads in %s" % out2
def main():
info('interleave-reads.py')
args = get_parser().parse_args()
for _ in args.infiles:
check_file_status(_, args.force)
check_space(args.infiles, args.force)
s1_file = args.infiles[0]
if len(args.infiles) == 2:
s2_file = args.infiles[1]
else:
s2_file = s1_file.replace('_R1_', '_R2_')
if s1_file == s2_file:
print >> sys.stderr, ("ERROR: given only one filename, that "
"doesn't contain _R1_. Exiting.")
sys.exit(1)
print >> sys.stderr, ("given only one file; "
"guessing that R2 file is %s" % s2_file)
fail = False
if not os.path.exists(s1_file):
print >> sys.stderr, "Error! R1 file %s does not exist" % s1_file
fail = True
if not os.path.exists(s2_file):
print >> sys.stderr, "Error! R2 file %s does not exist" % s2_file
fail = True
if fail and not args.force:
sys.exit(1)
print >> sys.stderr, "Interleaving:\n\t%s\n\t%s" % (s1_file, s2_file)
counter = 0
screed_iter_1 = screed.open(s1_file, parse_description=False)
screed_iter_2 = screed.open(s2_file, parse_description=False)
for read1, read2 in itertools.izip_longest(screed_iter_1, screed_iter_2):
if read1 is None or read2 is None:
print >> sys.stderr, ("ERROR: Input files contain different number"
" of records.")
sys.exit(1)
if counter % 100000 == 0:
print >> sys.stderr, '...', counter, 'pairs'
counter += 1
name1 = read1.name
if not check_is_left(name1):
name1 += '/1'
name2 = read2.name
if not check_is_right(name2):
name2 += '/2'
read1.name = name1
read2.name = name2
if not check_is_pair(read1, read2):
print >>sys.stderr, "ERROR: This doesn't look like paired data! " \
"%s %s" % (read1.name, read2.name)
sys.exit(1)
write_record_pair(read1, read2, args.output)
print >> sys.stderr, 'final: interleaved %d pairs' % counter
print >> sys.stderr, 'output written to', args.output.name
def main():
info('split-paired-reads.py')
args = get_parser().parse_args()
infile = args.infile
filenames = [infile]
check_input_files(infile, args.force)
check_space(filenames, args.force)
# decide where to put output files - specific directory? or just default?
if infile == '/dev/stdin' or infile == '-':
if not (args.output_first and args.output_second):
print >> sys.stderr, ("Accepting input from stdin; "
"output filenames must be provided.")
sys.exit(1)
elif args.output_directory:
if not os.path.exists(args.output_directory):
os.makedirs(args.output_directory)
out1 = args.output_directory + '/' + os.path.basename(infile) + '.1'
out2 = args.output_directory + '/' + os.path.basename(infile) + '.2'
else:
out1 = os.path.basename(infile) + '.1'
out2 = os.path.basename(infile) + '.2'
# OVERRIDE output file locations with -1, -2
if args.output_first:
fp_out1 = args.output_first
out1 = fp_out1.name
else:
# Use default filename created above
fp_out1 = open(out1, 'w')
if args.output_second:
fp_out2 = args.output_second
out2 = fp_out2.name
else:
# Use default filename created above
fp_out2 = open(out2, 'w')
counter1 = 0
counter2 = 0
index = None
screed_iter = screed.open(infile, parse_description=False)
# walk through all the reads in broken-paired mode.
paired_iter = broken_paired_reader(screed_iter)
for index, is_pair, record1, record2 in paired_iter:
if index % 10000 == 0:
print('...', index, file=sys.stderr)
# are we requiring pairs?
if args.force_paired and not is_pair:
print('ERROR, %s is not part of a pair' % record1.name,
file=sys.stderr)
sys.exit(1)
if is_pair:
write_record(record1, fp_out1)
counter1 += 1
write_record(record2, fp_out2)
counter2 += 1
else:
name = record1.name
if check_is_left(name):
write_record(record1, fp_out1)
counter1 += 1
elif check_is_right(name):
write_record(record1, fp_out2)
counter2 += 1
else:
print("Unrecognized format for read pair information: %s" %
name,
file=sys.stderr)
print("Exiting.", file=sys.stderr)
sys.exit(1)
print("DONE; split %d sequences (%d left, %d right)" %
(counter1 + counter2, counter1, counter2),
file=sys.stderr)
print("/1 reads in %s" % out1, file=sys.stderr)
print("/2 reads in %s" % out2, file=sys.stderr)
