def main(argv):
"""
Input EITHER (to generate per-transcript-results from pileup counts)
0. transcript/snp data
1. sample name {Col,Ped}
2. output prefix (output is <prefix>_<DATATYPE>.pickled)
3-n. number of bamfiles
OR (to summarise per-transcript-results)
0. name of a <arbitrary>_TRANSCRIPTDATA.pickled file
1. sample name {Col,Ped}
"""
if len(argv) > 3:
sys.stderr.write('%s: MODE1\n' % get_timestamp())
sys.stderr.write('')
transcript_d, snp_d = read_transcript_data(open(argv[0]))
# show_data(transcript_d, snp_d)
sample = argv[1]
prefix = argv[2]
read_checklist = set([])
for bam_fn in argv[3:]:
sys.stderr.write('%s: Processing file %s...\n' % (get_timestamp(), bam_fn))
process_pileups(pysam.Samfile(bam_fn, 'rb'), snp_d, read_checklist)
# show_data(transcript_d, snp_d)
# break
# print list(read_checklist), len(read_checklist)
ts = get_timestamp()
pickle.dump(snp_d, open(prefix + '_SNPDATA.pickled', 'wb'))
pickle.dump(transcript_d, open(prefix + '_TRANSCRIPTDATA.pickled', 'wb'))
pickle.dump(read_checklist, open(prefix + '_READCHECKLIST.pickled', 'wb'))
# sys.exit(0)
else:
sys.stderr.write('%s: MODE2\n' % get_timestamp())
sys.stderr.write('%s: Loading data from %s.\n' % (get_timestamp(), argv[0]))
# transcript_d = pickle.load(open(argv[0], 'rb'))
transcript_d = load_transcript_data(open(argv[0], 'rb'), sample=argv[1])
sys.stderr.write('%s: Finished loading %s.\n' % (get_timestamp(), argv[0]))
prefix = argv[0].rstrip('TRANSCRIPTDATA.pickled')
# print '======'
#show_data(transcript_d, None, sample=argv[1])
write_data(transcript_d, sample=argv[1], prefix=prefix)
pass
def main(argv):
sys.stderr.write('%s: Reading transcript/SNP data\n' % get_timestamp())
transcript_d, snp_d = read_transcript_data(open(argv[0]))
for bam_fn in argv[1:]:
sys.stderr.write('%s: Processing file %s...\n' % (get_timestamp(), bam_fn))
infile = pysam.Samfile(bam_fn, 'rb')
mapping_reads = get_mapping_reads(infile, snp_d)
bam_fn_out = bam_fn.replace('.bam', '.prefiltered.bam')
outfile = pysam.Samfile(bam_fn_out, 'wb', template=infile)
for read in infile.fetch():
if read.qname in mapping_reads:
outfile.write(read)
outfile.close()
infile.close()
pass
