def _xsq_convert_region(filename, sample, region, tags, outname):
out = gzip.open(outname, 'w')
xsq = XSQFile(filename)
for name, seq, quals in xsq.fetch_region(sample, region, tags):
if suffix:
out.write('@%s%s\n%s\n+\n%s\n' % (name, suffix, seq, ''.join([chr(q + 33) for q in quals])))
else:
out.write('@%s\n%s\n+\n%s\n' % (name, seq, ''.join([chr(q + 33) for q in quals])))
xsq.close()
out.close()
return region
def xsq_list(filename, count=False, minreads=-1, total=False):
xsq = XSQFile(filename)
print 'Tags: '
for tag in xsq.tags:
t = xsq.tags[tag]
if t.is_colorspace:
print ' %s[cs/%s]' % (tag, t.prefix)
else:
print ' %s[nt]' % (tag,)
print ''
print 'Samples: '
acc = 0
try:
for sample in xsq.get_samples():
desc = xsq.get_sample_desc(sample).strip()
if count:
readcount = xsq.get_read_count(sample)
if readcount > minreads:
pn = pretty_number(readcount)
if sample != 'Unclassified':
acc += readcount
if desc:
print ' %s (%s) %s' % (sample, desc, pn)
else:
print ' %s %s' % (sample, pn)
else:
if desc:
print ' %s (%s)' % (sample, desc)
else:
print ' %s' % (sample, )
except KeyboardInterrupt:
pass
if count and total:
print ''
print ' Total reads => %s' % pretty_number(acc)
print ''
xsq.close()
def xsq_convert_all(filename, tags=None, force=False, suffix=None, noz=False, usedesc=False, minreads=0, fsuffix=None, unclassified=False, procs=1, tmpdir=None):
xsq = XSQFile(filename)
samples = []
for sample in xsq.get_samples():
fname = sample
if not fsuffix:
fsuffix = ''
if usedesc:
fname = xsq.get_sample_desc(sample)
if not fname:
fname = sample
if fname == sample:
sys.stderr.write('Sample: %s... ' % fname)
else:
sys.stderr.write('Sample: (%s) %s... ' % (sample, fname))
if noz:
outname = '%s%s.fastq' % (fname, fsuffix)
else:
outname = '%s%s.fastq.gz' % (fname, fsuffix)
if force or not os.path.exists(outname):
if sample == 'Unclassified' and not unclassified:
sys.stderr.write(' Skipping unclassified\n')
continue
count = xsq.get_read_count(sample)
if count < minreads:
sys.stderr.write(' Too few reads (%s)\n' % count)
continue
samples.append((sample, outname))
sys.stderr.write('\n')
xsq.close()
for sample, outname in samples:
xsq_convert(filename, sample, tags, suffix, procs=procs, outname=outname, noz=noz, tmpdir=tmpdir)
def xsq_info(filename):
xsq = XSQFile(filename)
xsq.dump(xsq.hdf.root.RunMetadata)
xsq.close()
def xsq_convert(filename, sample=None, tags=None, suffix=None, procs=1, outname='-', tmpdir=None, noz=False):
sys.stderr.write("Converting: %s\n" % sample)
if tmpdir is None:
tmpdir = '.'
if procs < 1:
procs = multiprocessing.cpu_count()
pool = multiprocessing.Pool(procs)
xsq = XSQFile(filename)
regions = []
tmpnames = []
for region in xsq.get_regions(sample):
regions.append(region)
tmpnames.append(os.path.join(tmpdir, '.tmp.%s.%s.%s.fastq.gz.%s' % (os.path.basename(filename), sample, region, os.getpid())))
xsq.close()
if ETA:
callback = Callback(len(regions))
else:
callback = None
for region, tmpname in zip(regions, tmpnames):
pool.apply_async(_xsq_convert_region, (filename, sample, region, tags, tmpname), callback=callback)
pool.close()
try:
pool.join()
except KeyboardInterrupt:
pool.terminate()
sys.exit(1)
if callback:
callback.done()
sys.stderr.write("Merging temp files...\n")
if ETA:
callback = Callback(len(regions))
else:
callback = None
tmpname = os.path.join(tmpdir, '.tmp.%s.%s.%s' % (os.path.basename(outname), sample, os.getpid()))
if outname == '-':
out = sys.stdout
elif noz:
out = open(tmpname, 'w')
else:
out = gzip.open(tmpname, 'w')
for tmp in tmpnames:
src = gzip.open(tmp)
_dump_stream(src, out)
src.close()
os.unlink(tmp)
if callback:
callback()
if out != sys.stdout:
out.close()
shutil.move(tmpname, outname)
if callback:
callback.done()