def _run_somatic(paired, ref_file, target, out_file):
"""Run somatic calling with octopus, handling both paired and tumor-only cases.
TODO:
- Should we set downsample-above and downsample-target which default to 1000
and 500? How will these effect high depth panels and runtimes?
"""
align_bams = paired.tumor_bam
if paired.normal_bam:
align_bams += " %s --normal-sample %s" % (paired.normal_bam,
paired.normal_name)
cores = dd.get_num_cores(paired.tumor_data)
min_af = float(dd.get_min_allele_fraction(paired.tumor_data)) / 100.0
cmd = (
"octopus --threads {cores} --reference {ref_file} --reads {align_bams} "
"--regions-file {target} "
"--min-credible-somatic-frequency {min_af} "
"-C cancer "
"--working-directory {tmp_dir} "
"-o {tx_out_file} --legacy")
with file_transaction(paired.tumor_data, out_file) as tx_out_file:
tmp_dir = os.path.dirname(tx_out_file)
do.run(cmd.format(**locals()), "Octopus somatic calling")
_produce_compatible_vcf(tx_out_file, paired.tumor_data)
return out_file
def rnaseq_vardict_variant_calling(data):
sample = dd.get_sample_name(data)
variation_dir = os.path.join(dd.get_work_dir(data), "variation")
safe_makedir(variation_dir)
out_file = os.path.join(variation_dir, sample + "-vardict.vcf.gz")
if file_exists(out_file):
data = dd.set_vrn_file(data, out_file)
return data
vardict_cmd = vardict.get_vardict_command(data)
strandbias = "teststrandbias.R"
var2vcf = "var2vcf_valid.pl"
vcfstreamsort = config_utils.get_program("vcfstreamsort", data)
compress_cmd = "| bgzip -c"
freq = float(dd.get_min_allele_fraction(data, 20) / 100.0)
var2vcf_opts = "-v 50"
fix_ambig = vcfutils.fix_ambiguous_cl()
remove_dup = vcfutils.remove_dup_cl()
r_setup = ("unset R_HOME && export PATH=%s:$PATH && "
% os.path.dirname(Rscript_cmd()))
ref_file = dd.get_ref_file(data)
bamfile = dd.get_work_bam(data)
bed_file = gtf.gtf_to_bed(dd.get_gtf_file(data))
opts = " -c 1 -S 2 -E 3 -g 4 "
with file_transaction(out_file) as tx_out_file:
jvm_opts = vardict._get_jvm_opts(data, tx_out_file)
cmd = ("{r_setup}{jvm_opts}{vardict_cmd} -G {ref_file} -f {freq} "
"-N {sample} -b {bamfile} {opts} {bed_file} "
"| {strandbias}"
"| {var2vcf} -N {sample} -E -f {freq} {var2vcf_opts} "
"| {fix_ambig} | {remove_dup} | {vcfstreamsort} {compress_cmd} "
"> {tx_out_file}")
message = "Calling RNA-seq variants with VarDict"
do.run(cmd.format(**locals()), message)
data = dd.set_vrn_file(data, out_file)
return data
def _run_somatic(paired, ref_file, target, out_file):
"""Run somatic calling with octopus, handling both paired and tumor-only cases.
Tweaks for low frequency, tumor only and UMI calling documented in:
https://github.com/luntergroup/octopus/blob/develop/configs/UMI.config
"""
align_bams = paired.tumor_bam
if paired.normal_bam:
align_bams += " %s --normal-sample %s" % (paired.normal_bam, paired.normal_name)
cores = dd.get_num_cores(paired.tumor_data)
# Do not try to search below 0.4% currently as leads to long runtimes
# https://github.com/luntergroup/octopus/issues/29#issuecomment-428167979
min_af = max([float(dd.get_min_allele_fraction(paired.tumor_data)) / 100.0, 0.004])
min_af_floor = min_af / 4.0
cmd = ("octopus --threads {cores} --reference {ref_file} --reads {align_bams} "
"--regions-file {target} "
"--min-credible-somatic-frequency {min_af_floor} --min-expected-somatic-frequency {min_af} "
"--downsample-above 4000 --downsample-target 4000 --min-kmer-prune 5 --min-bubble-score 20 "
"--max-haplotypes 200 --somatic-snv-mutation-rate '5e-4' --somatic-indel-mutation-rate '1e-05' "
"--target-working-memory 5G --target-read-buffer-footprint 5G --max-somatic-haplotypes 3 "
"--caller cancer "
"--working-directory {tmp_dir} "
"-o {tx_out_file} --legacy")
if not paired.normal_bam:
cmd += (" --tumour-germline-concentration 5")
if dd.get_umi_type(paired.tumor_data) or _is_umi_consensus_bam(paired.tumor_bam):
cmd += (" --allow-octopus-duplicates --overlap-masking 0 "
"--somatic-filter-expression 'GQ < 200 | MQ < 30 | SB > 0.2 | SD[.25] > 0.1 "
"| BQ < 40 | DP < 100 | MF > 0.1 | AD < 5 | CC > 1.1 | GQD > 2'")
with file_transaction(paired.tumor_data, out_file) as tx_out_file:
tmp_dir = os.path.dirname(tx_out_file)
do.run(cmd.format(**locals()), "Octopus somatic calling")
_produce_compatible_vcf(tx_out_file, paired.tumor_data, is_somatic=True)
return out_file
def rnaseq_vardict_variant_calling(data):
sample = dd.get_sample_name(data)
out_dir = utils.safe_makedir(
os.path.join(dd.get_work_dir(data), "variation", "rnaseq", "vardict"))
out_file = os.path.join(out_dir, sample + "-vardict.vcf.gz")
if file_exists(out_file):
data = dd.set_vrn_file(data, out_file)
return data
vardict_cmd = vardict.get_vardict_command(data)
strandbias = "teststrandbias.R"
var2vcf = "var2vcf_valid.pl"
vcfstreamsort = config_utils.get_program("vcfstreamsort", data)
compress_cmd = "| bgzip -c"
freq = float(dd.get_min_allele_fraction(data, 20) / 100.0)
var2vcf_opts = "-v 50"
fix_ambig = vcfutils.fix_ambiguous_cl()
remove_dup = vcfutils.remove_dup_cl()
r_setup = get_R_exports()
ref_file = dd.get_ref_file(data)
bamfile = dd.get_work_bam(data)
data = _setup_variant_regions(data, out_dir)
opts, _ = vardict._vardict_options_from_config(
[data],
data["config"],
out_file,
dd.get_variant_regions(data),
is_rnaseq=True)
cores = dd.get_num_cores(data)
if cores and cores > 1:
opts += " -th %s" % str(cores)
with file_transaction(data, out_file) as tx_out_file:
jvm_opts = vardict._get_jvm_opts(data, tx_out_file)
cmd = ("{r_setup} && {jvm_opts}{vardict_cmd} -G {ref_file} -f {freq} "
"-N {sample} -b {bamfile} {opts} "
"| {strandbias}"
"| {var2vcf} -N {sample} -E -f {freq} {var2vcf_opts} "
"| {fix_ambig} | {remove_dup} | {vcfstreamsort} {compress_cmd} "
"> {tx_out_file}")
message = "Calling RNA-seq variants with VarDict"
do.run(cmd.format(**locals()), message)
out_file = vcfutils.bgzip_and_index(out_file, data["config"])
data = dd.set_vrn_file(data, out_file)
return data
def run(align_bams, items, ref_file, assoc_files, region=None, out_file=None):
"""Run tumor only pisces calling
Handles bgzipping output file and fixing VCF sample naming to match BAM sample.
"""
paired = vcfutils.get_paired_bams(align_bams, items)
assert paired and not paired.normal_bam, (
"Pisces supports tumor-only variant calling: %s" %
(",".join([dd.get_sample_name(d) for d in items])))
vrs = bedutils.population_variant_regions(items)
target = shared.subset_variant_regions(vrs,
region,
out_file,
items=items,
do_merge=True)
min_af = float(dd.get_min_allele_fraction(paired.tumor_data)) / 100.0
if not utils.file_exists(out_file):
base_out_name = utils.splitext_plus(os.path.basename(
paired.tumor_bam))[0]
raw_file = "%s.vcf" % utils.splitext_plus(out_file)[0]
with file_transaction(paired.tumor_data, raw_file) as tx_out_file:
ref_dir = _prep_genome(os.path.dirname(tx_out_file),
paired.tumor_data)
out_dir = os.path.dirname(tx_out_file)
cores = dd.get_num_cores(paired.tumor_data)
cmd = (
"pisces --bampaths {paired.tumor_bam} --genomepaths {ref_dir} --intervalpaths {target} "
"--maxthreads {cores} --minvf {min_af} --ploidy somatic --gvcf false -o {out_dir}"
)
do.run(cmd.format(**locals()), "Pisces tumor-only somatic calling")
shutil.move(os.path.join(out_dir, "%s.vcf" % base_out_name),
tx_out_file)
vcfutils.bgzip_and_index(
raw_file,
paired.tumor_data["config"],
prep_cmd="sed 's#%s.bam#%s#' | %s" %
(base_out_name, dd.get_sample_name(paired.tumor_data),
vcfutils.add_contig_to_header_cl(
dd.get_ref_file(paired.tumor_data), out_file)))
return vcfutils.bgzip_and_index(out_file, paired.tumor_data["config"])
def rnaseq_vardict_variant_calling(data):
sample = dd.get_sample_name(data)
out_dir = utils.safe_makedir(os.path.join(dd.get_work_dir(data),
"variation", "rnaseq", "vardict"))
out_file = os.path.join(out_dir, sample + "-vardict.vcf.gz")
if file_exists(out_file):
data = dd.set_vrn_file(data, out_file)
return data
vardict_cmd = vardict.get_vardict_command(data)
strandbias = "teststrandbias.R"
var2vcf = "var2vcf_valid.pl"
vcfstreamsort = config_utils.get_program("vcfstreamsort", data)
compress_cmd = "| bgzip -c"
freq = float(dd.get_min_allele_fraction(data, 20) / 100.0)
var2vcf_opts = "-v 50"
fix_ambig = vcfutils.fix_ambiguous_cl()
remove_dup = vcfutils.remove_dup_cl()
r_setup = get_R_exports()
ref_file = dd.get_ref_file(data)
bamfile = dd.get_work_bam(data)
data = _setup_variant_regions(data, out_dir)
opts, _ = vardict._vardict_options_from_config([data], data["config"], out_file, dd.get_variant_regions(data),
is_rnaseq=True)
cores = dd.get_num_cores(data)
if cores and cores > 1:
opts += " -th %s" % str(cores)
with file_transaction(data, out_file) as tx_out_file:
jvm_opts = vardict._get_jvm_opts(data, tx_out_file)
cmd = ("{r_setup} && {jvm_opts}{vardict_cmd} -G {ref_file} -f {freq} "
"-N {sample} -b {bamfile} {opts} "
"| {strandbias}"
"| {var2vcf} -N {sample} -E -f {freq} {var2vcf_opts} "
"| {fix_ambig} | {remove_dup} | {vcfstreamsort} {compress_cmd} "
"> {tx_out_file}")
message = "Calling RNA-seq variants with VarDict"
do.run(cmd.format(**locals()), message)
out_file = vcfutils.bgzip_and_index(out_file, data["config"])
data = dd.set_vrn_file(data, out_file)
return data
def run(align_bams, items, ref_file, assoc_files, region=None, out_file=None):
"""Run tumor only pisces calling
Handles bgzipping output file and fixing VCF sample naming to match BAM sample.
"""
paired = vcfutils.get_paired_bams(align_bams, items)
assert paired and not paired.normal_bam, ("Pisces supports tumor-only variant calling: %s" %
(",".join([dd.get_sample_name(d) for d in items])))
vrs = bedutils.population_variant_regions(items)
target = shared.subset_variant_regions(vrs, region,
out_file, items=items, do_merge=True)
min_af = float(dd.get_min_allele_fraction(paired.tumor_data)) / 100.0
if not utils.file_exists(out_file):
base_out_name = utils.splitext_plus(os.path.basename(paired.tumor_bam))[0]
raw_file = "%s.vcf" % utils.splitext_plus(out_file)[0]
with file_transaction(paired.tumor_data, raw_file) as tx_out_file:
ref_dir = _prep_genome(os.path.dirname(tx_out_file), paired.tumor_data)
out_dir = os.path.dirname(tx_out_file)
cores = dd.get_num_cores(paired.tumor_data)
emit_min_af = min_af / 10.0
cmd = ("pisces --bampaths {paired.tumor_bam} --genomepaths {ref_dir} --intervalpaths {target} "
"--maxthreads {cores} --minvf {emit_min_af} --vffilter {min_af} "
"--ploidy somatic --gvcf false -o {out_dir}")
# Recommended filtering for low frequency indels
# https://github.com/bcbio/bcbio-nextgen/commit/49d0cbb1f6dcbea629c63749e2f9813bd06dcee3#commitcomment-29765373
cmd += " -RMxNFilter 5,9,0.35"
# For low frequency UMI tagged variants, set higher variant thresholds
# https://github.com/Illumina/Pisces/issues/14#issuecomment-399756862
if min_af < (1.0 / 100.0):
cmd += " --minbasecallquality 30"
do.run(cmd.format(**locals()), "Pisces tumor-only somatic calling")
shutil.move(os.path.join(out_dir, "%s.vcf" % base_out_name),
tx_out_file)
vcfutils.bgzip_and_index(raw_file, paired.tumor_data["config"],
prep_cmd="sed 's#%s.bam#%s#' | %s" %
(base_out_name, dd.get_sample_name(paired.tumor_data),
vcfutils.add_contig_to_header_cl(dd.get_ref_file(paired.tumor_data), out_file)))
return vcfutils.bgzip_and_index(out_file, paired.tumor_data["config"])