def decode_fastq(infile, barcode2name, mismatch=0, startpos=37, outprefix=None):
"""
Decode barcodes in given fastq file
"""
if not outprefix:
outprefix = infile
barcode2outhandle = {
'failed':open(outprefix+'.failed.m'+str(mismatch), 'w'),
}
for bc in barcode2name:
outfile = outprefix+'.'+barcode2name[bc]+'.m'+str(mismatch)
barcode2outhandle[bc] = open(outfile, 'w')
for record in reader_fastq(infile):
try:
print record[1]
currseq = record[1][startpos-1:startpos+len(bc)-1]
except:
continue
barcode2dist = [(bc, barcode_distance(bc, currseq))
for bc in barcode2name]
#print >> sys.stderr, record[0], record[1], barcode2dist
min_dist = min([barcode2dist[i][1] for i in xrange(len(barcode2dist))])
bc_mindist = [barcode2dist[i][0] for i in xrange(len(barcode2dist))
if barcode2dist[i][1] == min_dist]
if min_dist > mismatch or len(bc_mindist) > 1:
barcode2outhandle['failed'].write('\n'.join(record)+'\n')
else:
barcode2outhandle[bc_mindist[0]].write('\n'.join(record)+'\n')
def decode(self):
"""decode and write results"""
for record in reader_fastq(self.infile):
fqseq = record.get_seq()
currseq = fqseq[self.startpos:(self.startpos+self.lenbc)]
if len(currseq) < self.lenbc:
sys.stderr.write("%s not have enough length" % currseq)
bc = self.decode_one(currseq)
self.barcode2outhandle[bc].write("%s\n" % str(record))
for oh in self.barcode2outhandle.values():
oh.close()