def main():
# read params
args = parse_arguments()
log.info('Initializing and making output directory...')
mkdir_p(args.out_dir)
# generate read length file
log.info('Generating read length file...')
make_read_length_file(args.fastq, args.out_dir)
log.info('Running samtools index...')
samtools_index(args.bam, args.nth, args.out_dir)
log.info('SAMstat on raw BAM...')
samstat(args.bam, args.nth, args.out_dir)
if args.chrsz:
log.info('SAMstat on non-mito BAM...')
non_mito_out_dir = os.path.join(args.out_dir, 'non_mito')
mkdir_p(non_mito_out_dir)
non_mito_bam = remove_chrs_from_bam(args.bam, [args.mito_chr_name],
args.chrsz, args.nth,
non_mito_out_dir)
samstat(non_mito_bam, args.nth, non_mito_out_dir)
rm_f(non_mito_bam)
log.info('List all files in output directory...')
ls_l(args.out_dir)
log.info('All done.')
def main():
# filt_bam - dupmark_bam - nodup_bam
# \ dup_qc \ pbc_qc
# read params
args = parse_arguments()
log.info('Initializing and making output directory...')
mkdir_p(args.out_dir)
# declare temp arrays
temp_files = [] # files to deleted later at the end
log.info('Removing unmapped/low-quality reads...')
if args.paired_end:
filt_bam = rm_unmapped_lowq_reads_pe(args.bam, args.multimapping,
args.mapq_thresh, args.nth,
args.mem_gb, args.out_dir)
else:
filt_bam = rm_unmapped_lowq_reads_se(args.bam, args.multimapping,
args.mapq_thresh, args.nth,
args.mem_gb, args.out_dir)
log.info('Checking if filtered BAM file is empty...')
if bam_is_empty(filt_bam, args.nth):
help_msg = (
'No reads found in filtered BAM. '
'Low quality sample? '
'Or no reads passing criteria "samtools view -F 1804"? '
'Check samtools flags at '
'https://broadinstitute.github.io/picard/explain-flags.html. ')
if args.paired_end:
help_msg += (
'Or is this truely PE BAM? '
'All unpaired SE reads could be removed by "samtools view -f 2". '
)
raise ValueError(help_msg)
log.info('Marking dupes with {}...'.format(args.dup_marker))
if args.dup_marker == 'picard':
dupmark_bam, dup_qc = mark_dup_picard(filt_bam, args.out_dir,
args.picard_java_heap)
elif args.dup_marker == 'sambamba':
dupmark_bam, dup_qc = mark_dup_sambamba(filt_bam, args.nth,
args.out_dir)
else:
raise argparse.ArgumentTypeError('Unsupported --dup-marker {}'.format(
args.dup_marker))
if args.no_dup_removal:
nodup_bam = filt_bam
else:
temp_files.append(filt_bam)
log.info('Removing dupes...')
if args.paired_end:
nodup_bam = rm_dup_pe(dupmark_bam, args.nth, args.out_dir)
else:
nodup_bam = rm_dup_se(dupmark_bam, args.nth, args.out_dir)
samtools_index(dupmark_bam)
temp_files.append(dupmark_bam + '.bai')
temp_files.append(dupmark_bam)
if len(args.filter_chrs) > 0:
final_bam = remove_chrs_from_bam(nodup_bam, args.filter_chrs,
args.chrsz, args.nth, args.out_dir)
temp_files.append(nodup_bam)
else:
final_bam = nodup_bam
log.info('Checking if final BAM file is empty...')
if bam_is_empty(final_bam, args.nth):
raise ValueError('No reads found in final (filtered/deduped) BAM. '
'Low quality sample? '
'Or BAM with duplicates only? ')
log.info('samtools index (final_bam)...')
samtools_index(final_bam, args.nth, args.out_dir)
log.info('samstat...')
samstat(final_bam, args.nth, args.mem_gb, args.out_dir)
log.info('Generating PBC QC log...')
if args.paired_end:
pbc_qc_pe(dupmark_bam, args.mito_chr_name, args.nth, args.out_dir)
else:
pbc_qc_se(dupmark_bam, args.mito_chr_name, args.out_dir)
log.info('samtools index (raw bam)...')
bam = copy_f_to_dir(args.bam, args.out_dir)
bai = samtools_index(bam, args.nth, args.out_dir)
temp_files.extend([bam, bai])
log.info('Removing temporary files...')
rm_f(temp_files)
log.info('List all files in output directory...')
ls_l(args.out_dir)
log.info('All done.')
def main():
# filt_bam - dupmark_bam - nodup_bam
# \ dup_qc \ pbc_qc
# read params
args = parse_arguments()
log.info('Initializing and making output directory...')
mkdir_p(args.out_dir)
# declare temp arrays
temp_files = [] # files to deleted later at the end
log.info('Removing unmapped/low-quality reads...')
if args.paired_end:
filt_bam = rm_unmapped_lowq_reads_pe(
args.bam, args.multimapping, args.mapq_thresh,
args.nth, args.out_dir)
else:
filt_bam = rm_unmapped_lowq_reads_se(
args.bam, args.multimapping, args.mapq_thresh,
args.nth, args.out_dir)
log.info('Marking dupes with {}...'.format(args.dup_marker))
if args.dup_marker == 'picard':
dupmark_bam, dup_qc = mark_dup_picard(
filt_bam, args.out_dir)
elif args.dup_marker == 'sambamba':
dupmark_bam, dup_qc = mark_dup_sambamba(
filt_bam, args.nth, args.out_dir)
else:
raise argparse.ArgumentTypeError(
'Unsupported --dup-marker {}'.format(args.dup_marker))
if args.no_dup_removal:
nodup_bam = filt_bam
else:
temp_files.append(filt_bam)
log.info('Removing dupes...')
if args.paired_end:
nodup_bam = rm_dup_pe(
dupmark_bam, args.nth, args.out_dir)
else:
nodup_bam = rm_dup_se(
dupmark_bam, args.nth, args.out_dir)
samtools_index(dupmark_bam)
temp_files.append(dupmark_bam+'.bai')
temp_files.append(dupmark_bam)
if len(args.filter_chrs) > 0:
final_bam = remove_chrs_from_bam(nodup_bam, args.filter_chrs,
args.chrsz, args.nth,
args.out_dir)
temp_files.append(nodup_bam)
else:
final_bam = nodup_bam
log.info('samtools index (final_bam)...')
samtools_index(final_bam, args.nth, args.out_dir)
log.info('samstat...')
samstat(final_bam, args.nth, args.out_dir)
log.info('Generating PBC QC log...')
if args.paired_end:
pbc_qc_pe(dupmark_bam, args.mito_chr_name, args.nth,
args.out_dir)
else:
pbc_qc_se(dupmark_bam, args.mito_chr_name, args.out_dir)
log.info('samtools index (raw bam)...')
bam = copy_f_to_dir(args.bam, args.out_dir)
bai = samtools_index(bam, args.nth, args.out_dir)
temp_files.extend([bam, bai])
log.info('Removing temporary files...')
rm_f(temp_files)
log.info('List all files in output directory...')
ls_l(args.out_dir)
log.info('All done.')