def convert_fastq(fq,ofq,out_lnum=4,out_baseQ=33,tickon = 10000):
nreads = preprocess_radtag_lane.get_read_count(fq)
lnum,baseQ = preprocess_radtag_lane.get_fastq_properties(fq)
fh = preprocess_radtag_lane.smartopen(fq)
ofh = preprocess_radtag_lane.smartopen(ofq,'w')
for i in xrange(nreads):
if i%tickon == 0:
print >> sys.stderr, '\r%s / %s (%0.1f%%)' % (i,nreads,(float(i)/nreads)*100),
n,s,qs = preprocess_radtag_lane.next_read_from_fh(fh, lnum)
ofh.write(preprocess_radtag_lane.as_fq_line(n,s,qs_to_q(qs,baseQ),out_baseQ,out_lnum))
print >> sys.stderr,'\n'
def convert_fastq(fq, ofq, out_lnum=4, out_baseQ=33, tickon=10000):
nreads = preprocess_radtag_lane.get_read_count(fq)
lnum, baseQ = preprocess_radtag_lane.get_fastq_properties(fq)
fh = preprocess_radtag_lane.smartopen(fq)
ofh = preprocess_radtag_lane.smartopen(ofq, 'w')
for i in xrange(nreads):
if i % tickon == 0:
print >> sys.stderr, '\r%s / %s (%0.1f%%)' % (i, nreads,
(float(i) / nreads) *
100),
n, s, qs = preprocess_radtag_lane.next_read_from_fh(fh, lnum)
ofh.write(
preprocess_radtag_lane.as_fq_line(n, s, qs_to_q(qs, baseQ),
out_baseQ, out_lnum))
print >> sys.stderr, '\n'
if len(sys.argv) == 4:
cutsite, fq, outfile = sys.argv[1:]
rc = preprocess_radtag_lane.get_read_count(fq)
lnum, baseQ = preprocess_radtag_lane.get_fastq_properties(fq)
fh = preprocess_radtag_lane.smartopen(fq)
ofh = preprocess_radtag_lane.smartopen(outfile, 'w')
found = 0
for i in range(rc):
if i > 0 and i % tick == 0:
print >> sys.stderr, '\r%s / %s (%0.1f%%) found %s (%0.1f%%)' % \
(i,rc,(float(i)/rc)*100,found,(float(found)/i)*100),
n, s, q = preprocess_radtag_lane.next_read_from_fh(fh, lnum)
if s[barcode_len:barcode_len + len(cutsite)] == cutsite:
line = preprocess_radtag_lane.as_fq_line(n, s, q, None, lnum)
ofh.write(line)
found += 1
ofh.close()
elif len(sys.argv) == 6:
cutsite, fq1, fq2, outfile1, outfile2 = sys.argv[1:]
rc1 = preprocess_radtag_lane.get_read_count(fq1)
rc2 = preprocess_radtag_lane.get_read_count(fq2)
if rc1 != rc2:
errstr = 'read count for %s = %s; %s = %s. counts must match' % (
fq1, rc1, fq2.rc2)
raise ValueError, errstr
lnum, baseQ = preprocess_radtag_lane.get_fastq_properties(fq1)
fh1 = preprocess_radtag_lane.smartopen(fq1)
ofh1 = preprocess_radtag_lane.smartopen(outfile1, 'w')
if len(sys.argv) == 4:
cutsite,fq,outfile = sys.argv[1:]
rc = preprocess_radtag_lane.get_read_count(fq)
lnum,baseQ = preprocess_radtag_lane.get_fastq_properties(fq)
fh = preprocess_radtag_lane.smartopen(fq)
ofh = preprocess_radtag_lane.smartopen(outfile,'w')
found = 0
for i in range(rc):
if i>0 and i % tick == 0:
print >> sys.stderr, '\r%s / %s (%0.1f%%) found %s (%0.1f%%)' % \
(i,rc,(float(i)/rc)*100,found,(float(found)/i)*100),
n,s,q = preprocess_radtag_lane.next_read_from_fh(fh,lnum)
if s[barcode_len:barcode_len+len(cutsite)] == cutsite:
line = preprocess_radtag_lane.as_fq_line(n,s,q,None,lnum)
ofh.write(line)
found += 1
ofh.close()
elif len(sys.argv) == 6:
cutsite,fq1,fq2,outfile1,outfile2 = sys.argv[1:]
rc1 = preprocess_radtag_lane.get_read_count(fq1)
rc2 = preprocess_radtag_lane.get_read_count(fq2)
if rc1 != rc2:
errstr = 'read count for %s = %s; %s = %s. counts must match' % (fq1,rc1,fq2.rc2)
raise ValueError, errstr
lnum,baseQ = preprocess_radtag_lane.get_fastq_properties(fq1)
fh1 = preprocess_radtag_lane.smartopen(fq1)
ofh1 = preprocess_radtag_lane.smartopen(outfile1,'w')
fh2 = preprocess_radtag_lane.smartopen(fq2)
