def filter_bam(bamfile, bedfile):
with pysam.Samfile(bamfile) as bam_handle_in:
regions = collect_regions(bedfile, bam_handle_in)
regions.reverse()
write_mode = samtools_compatible_wbu_mode()
with pysam.Samfile("-", write_mode,
template=bam_handle_in) as bam_handle_out:
while regions:
region_aend = 0
contig, start, end = regions[-1]
for record in bam_handle_in.fetch(contig, start):
current_aend = record.aend
region_aend = max(region_aend, current_aend)
if record.pos > end:
last_contig, _, _ = regions.pop()
if not regions:
break
contig, start, end = regions[-1]
if (region_aend + _BAM_BLOCK_SIZE < start) \
or (contig != last_contig):
break
if current_aend >= start:
bam_handle_out.write(record)
else: # Reached the end of this contig
while regions and (regions[-1][0] == contig):
regions.pop()
return 0
def setup_basic_batch(args, regions, prefix, func):
setup = {"files": {},
"temp_files": {},
"procs": {},
"handles": {}}
try:
setup["files"]["bed"] = write_bed_file(prefix, regions)
setup["temp_files"]["bed"] = setup["files"]["bed"]
setup["files"]["pipe"] = make_bam_pipe(prefix)
setup["temp_files"]["pipe"] = setup["files"]["pipe"]
setup["handles"]["outfile"] = open(prefix, "w")
zip_proc = popen(["bgzip"],
stdin=func(setup),
stdout=setup["handles"]["outfile"],
close_fds=True)
setup["procs"]["gzip"] = zip_proc
write_mode = samtools_compatible_wbu_mode()
setup["handles"]["bam_in"] = pysam.Samfile(args.bamfile)
setup["handles"]["bam_out"] = \
pysam.Samfile(setup["files"]["pipe"], write_mode,
template=setup["handles"]["bam_in"])
return setup
except:
traceback.print_exc()
cleanup_batch(setup)
raise
def setup_basic_batch(args, regions, prefix, func):
setup = {"files": {}, "temp_files": {}, "procs": {}, "handles": {}}
try:
setup["files"]["bed"] = write_bed_file(prefix, regions)
setup["temp_files"]["bed"] = setup["files"]["bed"]
setup["files"]["pipe"] = make_bam_pipe(prefix)
setup["temp_files"]["pipe"] = setup["files"]["pipe"]
setup["handles"]["outfile"] = open(prefix, "w")
zip_proc = popen(["bgzip"],
stdin=func(setup),
stdout=setup["handles"]["outfile"],
close_fds=True)
setup["procs"]["gzip"] = zip_proc
write_mode = samtools_compatible_wbu_mode()
setup["handles"]["bam_in"] = pysam.Samfile(args.bamfile)
setup["handles"]["bam_out"] = \
pysam.Samfile(setup["files"]["pipe"], write_mode,
template=setup["handles"]["bam_in"])
return setup
except:
traceback.print_exc()
cleanup_batch(setup)
raise
def filter_bam(bamfile, bedfile):
with pysam.Samfile(bamfile) as bam_handle_in:
regions = collect_regions(bedfile, bam_handle_in)
regions.reverse()
write_mode = samtools_compatible_wbu_mode()
with pysam.Samfile("-", write_mode, template=bam_handle_in) as bam_handle_out:
while regions:
region_aend = 0
contig, start, end = regions[-1]
for record in bam_handle_in.fetch(contig, start):
current_aend = record.aend
region_aend = max(region_aend, current_aend)
if record.pos > end:
last_contig, _, _ = regions.pop()
if not regions:
break
contig, start, end = regions[-1]
if (region_aend + _BAM_BLOCK_SIZE < start) or (contig != last_contig):
break
if current_aend >= start:
bam_handle_out.write(record)
else: # Reached the end of this contig
while regions and (regions[-1][0] == contig):
regions.pop()
return 0
def _pipe_to_bam():
"""Simply pipes a BAM/SAM file to stdout; this is required to handle SAM
files that do not contain records (i.e. only a header), which are not
properly handled by "samtools view -S -", resulting in a parse failure.
"""
with pysam.Samfile("-", "r") as input_handle:
write_mode = samtools_compatible_wbu_mode()
with pysam.Samfile("-", write_mode, template=input_handle) as output_handle:
for record in input_handle:
output_handle.write(record)
return 0
def _pipe_to_bam():
"""Simply pipes a BAM/SAM file to stdout; this is required to handle SAM
files that do not contain records (i.e. only a header), which are not
properly handled by "samtools view -S -", resulting in a parse failure.
"""
with pysam.Samfile("-", "r") as input_handle:
write_mode = samtools_compatible_wbu_mode()
with pysam.Samfile("-", write_mode, template=input_handle) as output_handle:
for record in input_handle:
output_handle.write(record)
return 0
def _cleanup_unmapped(args, cleanup_sam):
"""Reads a BAM (or SAM, if cleanup_sam is True) file from STDIN, and
filters reads according to the filters specified in the commandline
arguments 'args'. The resulting records are written to STDOUT in
uncompressed BAM format. The output BAM is marked as sorted (under the
assumption that 'samtools sort' is to be run on the output) and PG tags are
updated if specified in the args.
"""
spec = "r" if cleanup_sam else "rb"
with pysam.Samfile("-", spec) as input_handle:
header = copy.deepcopy(input_handle.header)
_set_sort_order(header)
_set_pg_tags(header, args.update_pg_tag)
if args.rg_id is not None:
_set_rg_tags(header, args.rg_id, args.rg)
write_mode = samtools_compatible_wbu_mode()
with pysam.Samfile("-", write_mode, header=header) as output_handle:
for record in input_handle:
if (record.mapq < args.min_quality) \
or (record.flag & args.exclude_flags):
continue
if record.is_unmapped:
# Unmapped read; clear all non-required fields
record = _cleanup_record(record)
elif record.mate_is_unmapped:
# Unmapped mate
record.rnext = record.tid
record.pnext = record.pos
record.tlen = 0
if args.rg_id is not None:
# Ensure that only one RG tag is set
tags = [(key, value) for (key, value) in record.tags
if key != "RG"]
tags.append(("RG", args.rg_id))
record.tags = tags
output_handle.write(record)
return 0
def _cleanup_unmapped(args, cleanup_sam):
"""Reads a BAM (or SAM, if cleanup_sam is True) file from STDIN, and
filters reads according to the filters specified in the commandline
arguments 'args'. The resulting records are written to STDOUT in
uncompressed BAM format. The output BAM is marked as sorted (under the
assumption that 'samtools sort' is to be run on the output) and PG tags are
updated if specified in the args.
"""
spec = "r" if cleanup_sam else "rb"
with pysam.Samfile("-", spec) as input_handle:
header = copy.deepcopy(input_handle.header)
_set_sort_order(header)
_set_pg_tags(header, args.update_pg_tag)
if args.rg_id is not None:
_set_rg_tags(header, args.rg_id, args.rg)
write_mode = samtools_compatible_wbu_mode()
with pysam.Samfile("-", write_mode, header=header) as output_handle:
for record in input_handle:
if (record.mapq < args.min_quality) \
or (record.flag & args.exclude_flags):
continue
if record.is_unmapped:
# Unmapped read; clear all non-required fields
record = _cleanup_record(record)
elif record.mate_is_unmapped:
# Unmapped mate
record.rnext = record.tid
record.pnext = record.pos
record.tlen = 0
if args.rg_id is not None:
# Ensure that only one RG tag is set
tags = [(key, value) for (key, value) in record.tags
if key != "RG"]
tags.append(("RG", args.rg_id))
record.tags = tags
output_handle.write(record)
return 0
