def main(argv):
fn_list = argv[0]
seqdir = argv[1]
seqfile = open(argv[2] + '.fa', 'wb')
setname = seqdir.split('_')[1]
sys.stderr.write('%s> Reading sets...\n' % get_timestamp())
seq_d = {'U': read_seqs(os.path.join(seqdir,
'%s_unpaired_all.fastq' % setname)),
'R1': read_seqs(os.path.join(seqdir,
'%s_R1_paired.fastq' % setname)),
'R2': read_seqs(os.path.join(seqdir,
'%s_R2_paired.fastq' % setname))}
sys.stderr.write('%s> Checking for duplicates...\n' % get_timestamp())
# seqfile = open(os.path.basename(fn_list) + '.fa', 'wb')
drf = DuplicateReadFinder()
for line in open(fn_list):
line = line.strip()
id_, sources = line.split('\t')
sources = sources.split(',')
for k in sources:
if len(sources) == 2:
kk = 'p'
elif sources[0].startswith('R'):
kk = 'u'
else:
kk = ''
if id_ in seq_d[k]:
drf.add_sequence(id_ + '_' + k + kk, seq_d[k][id_])
# seqfile.write('>%s_%s\n%s\n' % (id_, k, seq_d[k][id_]))
sys.stderr.write('%s> Writing sequences...\n' % get_timestamp())
drf.write_sequences(seqfile, modify_output=lambda x:x.replace('U', 'T'))
seqfile.close()
sys.stderr.write('%s> Done.\n' % get_timestamp())
pass
def main(argv):
fn_list = argv[0]
seqdir = argv[1]
seqfile = open(argv[2] + '.fa', 'wb')
drf = DuplicateReadFinder()
wanted = dict()
for line in open(fn_list):
line = line.strip()
id_, sources = line.split('\t')
sources = sources.split(',')
if len(sources) == 2:
kk = 'p'
elif sources[0].startswith('R'):
kk = 'u'
else:
kk = ''
for k in sources:
wanted[id_] = '%s_%s%s' % (id_, k, kk)
setname = seqdir.split('_')[1]
sys.stderr.write('%s> Reading sequences and checking for duplicates...\n' % get_timestamp())
for fn in ['%s_unpaired_all.fastq', '%s_R1_paired.fastq', '%s_R2_paired.fastq']:
read_seqs(os.path.join(seqdir, fn % setname), drf, wanted)
sys.stderr.write('%s> Writing sequences...\n' % get_timestamp())
drf.write_sequences(seqfile, modify_output=lambda x:x.replace('U', 'T'))
seqfile.close()
sys.stderr.write('%s> Done.\n' % get_timestamp())
pass
