def read_fastq(fastq_file,maxcnt):
gfr = GenericFastqFileReader(fastq_file)
ecnt = 0
qseen = set()
lenmax = 0
lenmin = float('inf')
entries = []
bases = 0
while True:
e = gfr.read_entry()
if not e or ecnt > maxcnt: break
ecnt += 1
slen = len(e['seq'])
if slen < lenmin: lenmin = slen
if slen > lenmax: lenmax = slen
seq = e['seq']
bases += len(seq)
for v in [ord(x) for x in e['quality']]:
qseen.add(v)
entries.append(e)
gfr.close()
qmin = min(qseen)
qmax = max(qseen)
stats = {}
stats['qmin'] = qmin
stats['qmax'] = qmax
stats['lenmin'] = lenmin
stats['lenmax'] = lenmax
stats['readcount'] = len(entries)
stats['basecount'] = bases
return [entries,stats]
def check_for_uniquely_named_reads(args):
observed_reads = set()
reads = {}
if args.fastq_reads:
gfr = GenericFastqFileReader(args.fastq_reads)
while True:
e = gfr.read_entry()
if not e: break
reads[e['name']] = e['seq']
if e['name'] in observed_reads:
sys.stderr.write("ERROR observed reads must be uniquely named")
sys.exit()
observed_reads.add(e['name'])
elif args.fasta_reads:
gfr = GenericFastaFileReader(args.fasta_reads)
while True:
e = gfr.read_entry()
if not e: break
reads[e['name']] = e['seq']
if e['name'] in observed_reads:
sys.stderr.write("ERROR observed reads must be uniquely named")
sys.exit()
observed_reads.add(e['name'])
return reads
def check_for_uniquely_named_reads(args):
observed_reads = set()
reads = {}
if args.fastq_reads:
gfr = GenericFastqFileReader(args.fastq_reads)
while True:
e = gfr.read_entry()
if not e: break
reads[e['name']] = e['seq']
if e['name'] in observed_reads:
sys.stderr.write("ERROR observed reads must be uniquely named")
sys.exit()
observed_reads.add(e['name'])
elif args.fasta_reads:
gfr = GenericFastaFileReader(args.fasta_reads)
while True:
e = gfr.read_entry()
if not e: break
reads[e['name']] = e['seq']
if e['name'] in observed_reads:
sys.stderr.write("ERROR observed reads must be uniquely named")
sys.exit()
observed_reads.add(e['name'])
return reads
