def __main__():
try:
fastqFile = sys.argv[1]
outFile = sys.argv[2]
subCnt = int(sys.argv[3]) # sub sample count
except:
print __doc__
sys.exit(-1)
## open the file
ffh = helper._open_file(fastqFile)
## counting number of reads in fastq file
read_cnt = 0
for rec in SeqIO.parse(ffh, 'fastq'):
read_cnt += 1
ffh.close()
print 'Number of reads in FASTQ: ', read_cnt
assert subCnt <= read_cnt, str(
subCnt
) + ' (sub-sample count) should be less than total read count ' + str(
read_cnt)
try:
accept_prob = (1.0 * subCnt) / read_cnt
except:
accept_prob = 1
## outfile directory check for creating the new file
try:
subFile = bz2.BZ2File(outFile, 'wb')
except Exception as error:
sys.exit(error)
cnt, sub_cnt = 0, 0
print 'Writing compressed file...'
ffh = helper._open_file(fastqFile)
for rec in SeqIO.parse(ffh, 'fastq'):
rnb = random.random()
cnt += 1
if rnb <= accept_prob:
sub_cnt += 1
subFile.write(rec.format("fastq"))
if subCnt == sub_cnt:
print '...done'
break
ffh.close()
subFile.close()
print 'Number of reads scanned: ', cnt
print 'Number of reads in: ', sub_cnt
def __main__():
try:
fastqFile = sys.argv[1]
outFile = sys.argv[2]
subCnt = int(sys.argv[3]) # sub sample count
except:
print __doc__
sys.exit(-1)
## open the file
ffh = helper._open_file(fastqFile)
## counting number of reads in fastq file
read_cnt = 0
for rec in SeqIO.parse(ffh, "fastq"):
read_cnt += 1
ffh.close()
print "Number of reads in FASTQ: ", read_cnt
assert subCnt <= read_cnt, str(subCnt) + " (sub-sample count) should be less than total read count " + str(read_cnt)
try:
accept_prob = (1.0 * subCnt) / read_cnt
except:
accept_prob = 1
## outfile directory check for creating the new file
try:
subFile = bz2.BZ2File(outFile, "wb")
except Exception as error:
sys.exit(error)
cnt, sub_cnt = 0, 0
print "Writing compressed file..."
ffh = helper._open_file(fastqFile)
for rec in SeqIO.parse(ffh, "fastq"):
rnb = random.random()
cnt += 1
if rnb <= accept_prob:
sub_cnt += 1
subFile.write(rec.format("fastq"))
if subCnt == sub_cnt:
print "...done"
break
ffh.close()
subFile.close()
print "Number of reads scanned: ", cnt
print "Number of reads in: ", sub_cnt
def __main__():
try:
fa_name = sys.argv[1]
except:
print __doc__
sys.exit(-1)
seq_info = dict()
fah = helper._open_file(fa_name)
for rec in SeqIO.parse(fah, "fasta"):
seq_info[rec.id] = len(rec.seq)
print rec.id, len(rec.seq)
fah.close()
print
print 'Number of FASTA entries: ', len(seq_info)
for long_one in sorted(seq_info.items(), key=itemgetter(1), reverse=True):
print 'Long contig length (bp): ', long_one[0], long_one[1]
break
for short_one in sorted(seq_info.items(), key=itemgetter(1)):
print 'Short contig length (bp): ', short_one[0], short_one[1]
break
flength = 0
for ele in sorted(seq_info.items(), key=itemgetter(1)):
flength += ele[1]
print 'Average length of FASTA contig (bp): ', (flength/len(seq_info))
print