def merge_postprocess_bams(inputs, output, tempdir, containers):
helpers.makedirs(tempdir)
merged_out = os.path.join(tempdir, 'merged_lanes.bam')
picardutils.merge_bams(inputs,
merged_out,
docker_image=containers['picard'])
bamutils.bam_index(merged_out,
merged_out + '.bai',
docker_image=containers['samtools'])
sorted_bam = os.path.join(tempdir, 'sorted.bam')
picardutils.bam_sort(merged_out,
sorted_bam,
tempdir,
docker_image=containers['picard'])
markdups_metrics = os.path.join(tempdir, 'markdups_metrics.txt')
picardutils.bam_markdups(sorted_bam,
output,
markdups_metrics,
tempdir,
docker_image=containers['picard'])
bamutils.bam_index(output,
output + '.bai',
docker_image=containers['samtools'])
def merge_bams(inputs, output, output_index, config):
container_ctx = helpers.get_container_ctx(config['containers'],
'picard',
docker_only=True)
picardutils.merge_bams(inputs, output, **container_ctx)
container_ctx = helpers.get_container_ctx(config['containers'],
'samtools',
docker_only=True)
bamutils.bam_index(output, output_index, **container_ctx)
def cell_region_merge_bams(cell_bams, region_bam, region):
cell_bams = cell_bams.values()
region = '{}:{}-{}'.format(*region.split('-'))
bamutils.bam_merge(cell_bams, region_bam, region=region)
bamutils.bam_index(
region_bam,
region_bam + '.bai',
)
def merge_postprocess_bams(inputs, output, tempdir):
helpers.makedirs(tempdir)
merged_out = os.path.join(tempdir, 'merged_lanes.bam')
picardutils.merge_bams(inputs, merged_out)
bamutils.bam_index(merged_out, merged_out + '.bai')
sorted_bam = os.path.join(tempdir, 'sorted.bam')
picardutils.bam_sort(merged_out, sorted_bam, tempdir)
markdups_metrics = os.path.join(tempdir, 'markdups_metrics.txt')
picardutils.bam_markdups(sorted_bam, output, markdups_metrics, tempdir)
bamutils.bam_index(output, output + '.bai')
def postprocess_bam(infile, outfile, tempdir, containers):
outfile_index = outfile + '.bai'
if not os.path.exists(tempdir):
helpers.makedirs(tempdir)
sorted_bam = os.path.join(tempdir, 'sorted.bam')
picardutils.bam_sort(infile, sorted_bam, tempdir,
docker_image=containers['picard'])
markdups_metrics = os.path.join(tempdir, 'markdups_metrics.txt')
picardutils.bam_markdups(sorted_bam, outfile, markdups_metrics, tempdir,
docker_image=containers['picard'])
bamutils.bam_index(outfile, outfile_index, docker_image=containers['samtools'])
def remove_softclipped_reads(infile, outfile, softclipped_reads_threshold):
if softclipped_reads_threshold == 1:
shutil.copyfile(infile, outfile)
shutil.copyfile(infile + '.bai', outfile + '.bai')
return
bamfile = pysam.AlignmentFile(infile, "rb")
filteredbam = pysam.AlignmentFile(outfile, "wb", template=bamfile)
for read in bamfile.fetch():
if _fraction_softclipped(read) < softclipped_reads_threshold:
filteredbam.write(read)
filteredbam.close()
bam_index(outfile, outfile + '.bai')
def postprocess_bam(infile, outfile, outfile_index, tempdir, config,
markdups_metrics, flagstat_metrics):
if not os.path.exists(tempdir):
helpers.makedirs(tempdir)
container_ctx = helpers.get_container_ctx(config['containers'],
'picard',
docker_only=True)
sorted_bam = os.path.join(tempdir, 'sorted.bam')
picardutils.bam_sort(infile, sorted_bam, tempdir, **container_ctx)
picardutils.bam_markdups(sorted_bam, outfile, markdups_metrics, tempdir,
**container_ctx)
container_ctx = helpers.get_container_ctx(config['containers'],
'samtools',
docker_only=True)
bamutils.bam_index(outfile, outfile_index, **container_ctx)
bamutils.bam_flagstat(outfile, flagstat_metrics, **container_ctx)
def get_postprocess_metrics(infile, infile_bai, tempdir, config,
markdups_metrics, flagstat_metrics):
if not os.path.exists(tempdir):
helpers.makedirs(tempdir)
outfile = os.path.join(tempdir, 'markdps.bam')
outfile_index = outfile + '.bai'
container_ctx = helpers.get_container_ctx(config['containers'],
'picard',
docker_only=True)
picardutils.bam_markdups(infile, outfile, markdups_metrics, tempdir,
**container_ctx)
container_ctx = helpers.get_container_ctx(config['containers'],
'samtools',
docker_only=True)
bamutils.bam_index(outfile, outfile_index, **container_ctx)
bamutils.bam_flagstat(outfile, flagstat_metrics, **container_ctx)
def split_bam_file(bam, outbam, interval, samtools_docker):
outbai = outbam + '.bai'
bamutils.bam_view(bam, outbam, interval, docker_image=samtools_docker)
bamutils.bam_index(outbam, outbai, docker_image=samtools_docker)
def merge_bam_worker(input_bam_files, output_bam, region, kwargs):
bamutils.bam_merge(input_bam_files, output_bam, region=region, **kwargs)
bamutils.bam_index(output_bam, output_bam + '.bai', **kwargs)
def split_bam_file(bam, bai, outbam, outbai, interval, kwargs):
bamutils.bam_view(bam, outbam, interval, **kwargs)
bamutils.bam_index(outbam, outbai, **kwargs)
def index_bam_worker(bam, kwargs):
bamutils.bam_index(
bam, bam+'.bai', **kwargs)
def merge_bams(inputs, output, output_index, containers):
picardutils.merge_bams(inputs, output, docker_image=containers['picard'])
bamutils.bam_index(output, output_index, docker_image=containers['samtools'])
def split_bam_file(bam, outbam, interval):
outbai = outbam + '.bai'
bamutils.bam_view(bam, outbam, interval)
bamutils.bam_index(outbam, outbai)