def combinePaired(outdir,
ref,
tbam,
nbam,
inclusion=None,
exclusion=None,
mutect=None,
indelocator=None,
mutect2=None,
varscan_snv=None,
varscan_indel=None,
jsm=None,
sniper=None,
vardict=None,
muse=None,
lofreq_snv=None,
lofreq_indel=None,
scalpel=None,
strelka_snv=None,
strelka_indel=None,
tnscope=None,
platypus=None,
keep_intermediates=False):
hg_dict = re.sub(r'\.fa(sta)?$', '.dict', ref)
intermediate_files = set()
snv_intermediates = []
indel_intermediates = []
intermediate_vcfs = {'MuTect2': {'snv': None, 'indel': None}, \
'VarDict': {'snv': None, 'indel': None}, \
'TNscope': {'snv': None, 'indel': None}, \
'Platypus': {'snv': None, 'indel': None} }
# Modify direct VCF outputs for merging:
if mutect or indelocator:
import vcfModifier.modify_MuTect as mod_mutect
if mutect:
mutect_in = bed_intersector(
mutect, os.sep.join((outdir, 'intersect.mutect1.vcf')),
inclusion, exclusion)
intermediate_files.add(mutect_in)
snv_mutect_out = os.sep.join((outdir, 'snv.mutect1.vcf'))
mod_mutect.convert(mutect_in, snv_mutect_out, tbam, nbam)
intermediate_files.add(snv_mutect_out)
snv_intermediates.append(snv_mutect_out)
if indelocator:
indelocator_in = bed_intersector(
indelocator, os.sep.join(
(outdir, 'intersect.indelocator.vcf')), inclusion,
exclusion)
intermediate_files.add(indelocator_in)
indel_indelocator_out = os.sep.join(
(outdir, 'indel.indelocator.vcf'))
mod_mutect.convert(indelocator_in, indel_indelocator_out, tbam,
nbam)
intermediate_files.add(indel_indelocator_out)
indel_intermediates.append(indel_indelocator_out)
if mutect2:
import vcfModifier.modify_MuTect2 as mod_mutect2
mutect2_in = bed_intersector(
mutect2, os.sep.join((outdir, 'intersect.mutect2.vcf')), inclusion,
exclusion)
intermediate_files.add(mutect2_in)
snv_mutect_out = os.sep.join((outdir, 'snv.mutect2.vcf'))
indel_mutect_out = os.sep.join((outdir, 'indel.mutect2.vcf'))
mod_mutect2.convert(mutect2_in, snv_mutect_out, indel_mutect_out,
False)
for file_i in snv_mutect_out, indel_mutect_out:
intermediate_files.add(file_i)
snv_intermediates.append(snv_mutect_out)
indel_intermediates.append(indel_mutect_out)
intermediate_vcfs['MuTect2']['snv'] = snv_mutect_out
intermediate_vcfs['MuTect2']['indel'] = indel_mutect_out
if varscan_snv or varscan_indel:
import vcfModifier.modify_VarScan2 as mod_varscan2
if varscan_snv:
varscan_in = bed_intersector(
varscan_snv, os.sep.join(
(outdir, 'intersect.varscan.snv.vcf')), inclusion,
exclusion)
intermediate_files.add(varscan_in)
snv_varscan_out = os.sep.join((outdir, 'snv.varscan.vcf'))
mod_varscan2.convert(varscan_in, snv_varscan_out)
intermediate_files.add(snv_varscan_out)
snv_intermediates.append(snv_varscan_out)
if varscan_indel:
varscan_in = bed_intersector(
varscan_indel,
os.sep.join((outdir, 'intersect.varscan.indel.vcf')),
inclusion, exclusion)
intermediate_files.add(varscan_in)
indel_varscan_out = os.sep.join((outdir, 'indel.varscan.vcf'))
mod_varscan2.convert(varscan_in, indel_varscan_out)
intermediate_files.add(indel_varscan_out)
indel_intermediates.append(indel_varscan_out)
if jsm:
import vcfModifier.modify_JointSNVMix2 as mod_jsm
jsm_in = bed_intersector(jsm, os.sep.join(
(outdir, 'intersect.jsm.vcf')), inclusion, exclusion)
intermediate_files.add(jsm_in)
jsm_out = os.sep.join((outdir, 'snv.jsm.vcf'))
mod_jsm.convert(jsm_in, jsm_out)
intermediate_files.add(jsm_out)
snv_intermediates.append(jsm_out)
if sniper:
import vcfModifier.modify_SomaticSniper as mod_sniper
sniper_in = bed_intersector(
sniper, os.sep.join((outdir, 'intersect.sniper.vcf')), inclusion,
exclusion)
intermediate_files.add(sniper_in)
sniper_out = os.sep.join((outdir, 'snv.somaticsniper.vcf'))
mod_sniper.convert(sniper_in, sniper_out)
intermediate_files.add(sniper_out)
snv_intermediates.append(sniper_out)
if vardict:
import vcfModifier.modify_VarDict as mod_vardict
vardict_in = bed_intersector(
vardict, os.sep.join((outdir, 'intersect.vardict.vcf')), inclusion,
exclusion)
intermediate_files.add(vardict_in)
snv_vardict_out = os.sep.join((outdir, 'snv.vardict.vcf'))
indel_vardict_out = os.sep.join((outdir, 'indel.vardict.vcf'))
mod_vardict.convert(vardict_in, snv_vardict_out, indel_vardict_out)
sorted_snv_vardict_out = os.sep.join((outdir, 'snv.sort.vardict.vcf'))
sorted_indel_vardict_out = os.sep.join(
(outdir, 'indel.sort.vardict.vcf'))
vcfsorter(ref, snv_vardict_out, sorted_snv_vardict_out)
vcfsorter(ref, indel_vardict_out, sorted_indel_vardict_out)
for file_i in snv_vardict_out, indel_vardict_out, sorted_snv_vardict_out, sorted_indel_vardict_out:
intermediate_files.add(file_i)
snv_intermediates.append(sorted_snv_vardict_out)
indel_intermediates.append(sorted_indel_vardict_out)
intermediate_vcfs['VarDict']['snv'] = sorted_snv_vardict_out
intermediate_vcfs['VarDict']['indel'] = sorted_indel_vardict_out
if muse:
muse_in = bed_intersector(muse,
os.sep.join((outdir, 'intersect.muse.vcf')),
inclusion, exclusion)
intermediate_files.add(muse_in)
muse_out = os.sep.join((outdir, 'snv.muse.vcf'))
copy_TextFile.copy(muse_in, muse_out)
intermediate_files.add(muse_out)
snv_intermediates.append(muse_out)
if lofreq_snv:
lofreq_in = bed_intersector(
lofreq_snv, os.sep.join((outdir, 'intersect.lofreq.snv.vcf')),
inclusion, exclusion)
intermediate_files.add(lofreq_in)
snv_lofreq_out = os.sep.join((outdir, 'snv.lofreq.vcf'))
copy_TextFile.copy(lofreq_in, snv_lofreq_out)
intermediate_files.add(snv_lofreq_out)
snv_intermediates.append(snv_lofreq_out)
if lofreq_indel:
lofreq_in = bed_intersector(
lofreq_indel, os.sep.join((outdir, 'intersect.lofreq.indel.vcf')),
inclusion, exclusion)
intermediate_files.add(lofreq_in)
indel_lofreq_out = os.sep.join((outdir, 'indel.lofreq.vcf'))
copy_TextFile.copy(lofreq_in, indel_lofreq_out)
intermediate_files.add(indel_lofreq_out)
indel_intermediates.append(indel_lofreq_out)
if scalpel:
scalpel_in = bed_intersector(
scalpel, os.sep.join((outdir, 'intersect.scalpel.vcf')), inclusion,
exclusion)
intermediate_files.add(scalpel_in)
scalpel_out = os.sep.join((outdir, 'indel.scalpel.vcf'))
copy_TextFile.copy(scalpel_in, scalpel_out)
intermediate_files.add(scalpel_out)
indel_intermediates.append(scalpel_out)
if strelka_snv or strelka_indel:
import vcfModifier.modify_Strelka as mod_strelka
if strelka_snv:
strelka_in = bed_intersector(
strelka_snv, os.sep.join(
(outdir, 'intersect.strelka.snv.vcf')), inclusion,
exclusion)
intermediate_files.add(strelka_in)
snv_strelka_out = os.sep.join((outdir, 'snv.strelka.vcf'))
mod_strelka.convert(strelka_in, snv_strelka_out)
intermediate_files.add(snv_strelka_out)
snv_intermediates.append(snv_strelka_out)
if strelka_indel:
strelka_in = bed_intersector(
strelka_indel,
os.sep.join((outdir, 'intersect.strelka.indel.vcf')),
inclusion, exclusion)
intermediate_files.add(strelka_in)
indel_strelka_out = os.sep.join((outdir, 'indel.strelka.vcf'))
mod_strelka.convert(strelka_in, indel_strelka_out)
intermediate_files.add(indel_strelka_out)
indel_intermediates.append(indel_strelka_out)
if tnscope:
import vcfModifier.modify_MuTect2 as mod_mutect2
tnscope_in = bed_intersector(
tnscope, os.sep.join((outdir, 'intersect.tnscope.vcf')), inclusion,
exclusion)
intermediate_files.add(tnscope_in)
snv_tnscope_out = os.sep.join((outdir, 'snv.tnscope.vcf'))
indel_tnscope_out = os.sep.join((outdir, 'indel.tnscope.vcf'))
mod_mutect2.convert(tnscope_in, snv_tnscope_out, indel_tnscope_out,
True)
for file_i in snv_tnscope_out, indel_tnscope_out:
intermediate_files.add(file_i)
snv_intermediates.append(snv_tnscope_out)
indel_intermediates.append(indel_tnscope_out)
intermediate_vcfs['TNscope']['snv'] = snv_tnscope_out
intermediate_vcfs['TNscope']['indel'] = indel_tnscope_out
if platypus:
platypus_in = bed_intersector(
platypus, os.sep.join((outdir, 'intersect.platypus.vcf')),
inclusion, exclusion)
intermediate_files.add(platypus_in)
snv_platypus_out = os.sep.join((outdir, 'snv.platypus.vcf'))
indel_platypus_out = os.sep.join((outdir, 'indel.platypus.vcf'))
splitVcf.split_into_snv_and_indel(platypus_in, snv_platypus_out,
indel_platypus_out)
for file_i in snv_platypus_out, indel_platypus_out:
intermediate_files.add(file_i)
snv_intermediates.append(snv_platypus_out)
indel_intermediates.append(indel_platypus_out)
intermediate_vcfs['Platypus']['snv'] = snv_platypus_out
intermediate_vcfs['Platypus']['indel'] = indel_platypus_out
# Combine SNV/INDEL variant candidates
snv_combined = os.sep.join((outdir, 'unsorted.CombineVariants.snv.vcf'))
indel_combined = os.sep.join(
(outdir, 'unsorted.CombineVariants.indel.vcf'))
getUniqueVcfPositions.combine(snv_intermediates, snv_combined)
getUniqueVcfPositions.combine(indel_intermediates, indel_combined)
for file_i in snv_combined, indel_combined:
intermediate_files.add(file_i)
# Sort them:
snv_combined_sorted = os.sep.join((outdir, 'CombineVariants.snv.vcf'))
indel_combined_sorted = os.sep.join((outdir, 'CombineVariants.indel.vcf'))
vcfsorter(ref, snv_combined, snv_combined_sorted)
vcfsorter(ref, indel_combined, indel_combined_sorted)
if not keep_intermediates:
for file_i in intermediate_files:
subprocess.call(('rm', '-v', file_i))
return snv_combined_sorted, indel_combined_sorted, intermediate_vcfs, intermediate_files
def combineSingle(outdir,
ref,
bam,
inclusion=None,
exclusion=None,
mutect=None,
mutect2=None,
varscan=None,
vardict=None,
lofreq=None,
scalpel=None,
strelka=None,
keep_intermediates=False):
hg_dict = re.sub(r'\.fa(sta)?$', '.dict', ref)
intermediate_files = set()
snv_intermediates = []
indel_intermediates = []
intermediate_vcfs = {'MuTect2' :{'snv': None, 'indel': None}, \
'VarScan2' :{'snv': None, 'indel': None}, \
'VarDict' :{'snv': None, 'indel': None}, \
'LoFreq' :{'snv': None, 'indel': None}, \
'Strelka' :{'snv': None, 'indel': None}, }
if mutect:
import vcfModifier.modify_MuTect as mod_mutect
mutect_in = bed_intersector(
mutect, os.sep.join((outdir, 'intersect.mutect1.vcf')), inclusion,
exclusion)
intermediate_files.add(mutect_in)
snv_mutect_out = os.sep.join((outdir, 'snv.mutect1.vcf'))
mod_mutect.convert(mutect_in, snv_mutect_out, bam)
intermediate_files.add(snv_mutect_out)
snv_intermediates.append(snv_mutect_out)
if mutect2:
import vcfModifier.modify_ssMuTect2 as mod_mutect2
mutect2_in = bed_intersector(
mutect2, os.sep.join((outdir, 'intersect.mutect2.vcf')), inclusion,
exclusion)
intermediate_files.add(mutect2_in)
snv_mutect_out = os.sep.join((outdir, 'snv.mutect2.vcf'))
indel_mutect_out = os.sep.join((outdir, 'indel.mutect2.vcf'))
mod_mutect2.convert(mutect2_in, snv_mutect_out, indel_mutect_out)
for file_i in snv_mutect_out, indel_mutect_out:
intermediate_files.add(file_i)
snv_intermediates.append(snv_mutect_out)
indel_intermediates.append(indel_mutect_out)
intermediate_vcfs['MuTect2']['snv'] = snv_mutect_out
intermediate_vcfs['MuTect2']['indel'] = indel_mutect_out
if varscan:
import vcfModifier.modify_VarScan2 as mod_varscan2
varscan_in = bed_intersector(
varscan, os.sep.join((outdir, 'intersect.varscan.vcf')), inclusion,
exclusion)
intermediate_files.add(varscan_in)
snv_temp = os.sep.join((outdir, 'snv.varscan.temp.vcf'))
indel_temp = os.sep.join((outdir, 'indel.varscan.temp.vcf'))
snv_varscan_out = os.sep.join((outdir, 'snv.varscan.vcf'))
indel_varscan_out = os.sep.join((outdir, 'indel.varscan.vcf'))
splitVcf.split_into_snv_and_indel(varscan_in, snv_temp, indel_temp)
mod_varscan2.convert(snv_temp, snv_varscan_out)
mod_varscan2.convert(indel_temp, indel_varscan_out)
for file_i in snv_temp, indel_temp, snv_varscan_out, indel_varscan_out:
intermediate_files.add(file_i)
snv_intermediates.append(snv_varscan_out)
indel_intermediates.append(indel_varscan_out)
intermediate_vcfs['VarScan2']['snv'] = snv_varscan_out
intermediate_vcfs['VarScan2']['indel'] = indel_varscan_out
if vardict:
import vcfModifier.modify_VarDict as mod_vardict
vardict_in = bed_intersector(
vardict, os.sep.join((outdir, 'intersect.vardict.vcf')), inclusion,
exclusion)
intermediate_files.add(vardict_in)
snv_vardict_out = os.sep.join((outdir, 'snv.vardict.vcf'))
indel_vardict_out = os.sep.join((outdir, 'indel.vardict.vcf'))
mod_vardict.convert(vardict_in, snv_vardict_out, indel_vardict_out)
sorted_snv_vardict_out = os.sep.join((outdir, 'snv.sort.vardict.vcf'))
sorted_indel_vardict_out = os.sep.join(
(outdir, 'indel.sort.vardict.vcf'))
vcfsorter(ref, snv_vardict_out, sorted_snv_vardict_out)
vcfsorter(ref, indel_vardict_out, sorted_indel_vardict_out)
for file_i in snv_vardict_out, indel_vardict_out, sorted_snv_vardict_out, sorted_indel_vardict_out:
intermediate_files.add(file_i)
snv_intermediates.append(sorted_snv_vardict_out)
indel_intermediates.append(sorted_indel_vardict_out)
intermediate_vcfs['VarDict']['snv'] = sorted_snv_vardict_out
intermediate_vcfs['VarDict']['indel'] = sorted_indel_vardict_out
if lofreq:
lofreq_in = bed_intersector(
lofreq, os.sep.join((outdir, 'intersect.lofreq.vcf')), inclusion,
exclusion)
intermediate_files.add(lofreq_in)
snv_lofreq_out = os.sep.join((outdir, 'snv.lofreq.vcf'))
indel_lofreq_out = os.sep.join((outdir, 'indel.lofreq.vcf'))
splitVcf.split_into_snv_and_indel(lofreq_in, snv_lofreq_out,
indel_lofreq_out)
for file_i in snv_lofreq_out, indel_lofreq_out:
intermediate_files.add(file_i)
snv_intermediates.append(snv_lofreq_out)
indel_intermediates.append(indel_lofreq_out)
intermediate_vcfs['LoFreq']['snv'] = snv_lofreq_out
intermediate_vcfs['LoFreq']['indel'] = indel_lofreq_out
if scalpel:
scalpel_in = bed_intersector(
scalpel, os.sep.join((outdir, 'intersect.scalpel.vcf')), inclusion,
exclusion)
intermediate_files.add(scalpel_in)
scalpel_out = os.sep.join((outdir, 'indel.scalpel.vcf'))
copy_TextFile.copy(scalpel_in, scalpel_out)
intermediate_files.add(scalpel_out)
indel_intermediates.append(scalpel_out)
if strelka:
import vcfModifier.modify_ssStrelka as mod_strelka
strelka_in = bed_intersector(
strelka, os.sep.join((outdir, 'intersect.strelka.vcf')), inclusion,
exclusion)
intermediate_files.add(strelka_in)
snv_strelka_out = os.sep.join((outdir, 'snv.strelka.vcf'))
indel_strelka_out = os.sep.join((outdir, 'indel.strelka.vcf'))
mod_strelka.convert(strelka_in, snv_strelka_out, indel_strelka_out)
for file_i in snv_strelka_out, indel_strelka_out:
intermediate_files.add(file_i)
snv_intermediates.append(snv_strelka_out)
indel_intermediates.append(indel_strelka_out)
intermediate_vcfs['Strelka']['snv'] = snv_strelka_out
intermediate_vcfs['Strelka']['indel'] = indel_strelka_out
# Combine SNV/INDEL variant candidates
snv_combined = os.sep.join((outdir, 'unsorted.CombineVariants.snv.vcf'))
indel_combined = os.sep.join(
(outdir, 'unsorted.CombineVariants.indel.vcf'))
getUniqueVcfPositions.combine(snv_intermediates, snv_combined)
getUniqueVcfPositions.combine(indel_intermediates, indel_combined)
for file_i in snv_combined, indel_combined:
intermediate_files.add(file_i)
# Sort them:
snv_combined_sorted = os.sep.join((outdir, 'CombineVariants.snv.vcf'))
indel_combined_sorted = os.sep.join((outdir, 'CombineVariants.indel.vcf'))
vcfsorter(ref, snv_combined, snv_combined_sorted)
vcfsorter(ref, indel_combined, indel_combined_sorted)
if not keep_intermediates:
for file_i in intermediate_files:
subprocess.call(('rm', '-v', file_i))
return snv_combined_sorted, indel_combined_sorted, intermediate_vcfs, intermediate_files
def combineSingle(outdir, ref, bam, inclusion=None, exclusion=None, mutect=None, mutect2=None, varscan=None, vardict=None, lofreq=None, scalpel=None, strelka=None, keep_intermediates=False):
hg_dict = re.sub(r'\.fa(sta)?$', '.dict', ref)
intermediate_files = set()
snv_intermediates = []
indel_intermediates = []
intermediate_vcfs = {'MuTect2' :{'snv': None, 'indel': None}, \
'VarScan2' :{'snv': None, 'indel': None}, \
'VarDict' :{'snv': None, 'indel': None}, \
'LoFreq' :{'snv': None, 'indel': None}, \
'Strelka' :{'snv': None, 'indel': None}, }
if mutect:
import vcfModifier.modify_MuTect as mod_mutect
mutect_in = bed_intersector(mutect, os.sep.join(( outdir, 'intersect.mutect1.vcf' )), inclusion, exclusion)
intermediate_files.add(mutect_in)
snv_mutect_out = os.sep.join(( outdir, 'snv.mutect1.vcf' ))
mod_mutect.convert(mutect_in, snv_mutect_out, bam)
intermediate_files.add(snv_mutect_out)
snv_intermediates.append(snv_mutect_out)
if mutect2:
import vcfModifier.modify_ssMuTect2 as mod_mutect2
mutect2_in = bed_intersector(mutect2, os.sep.join(( outdir, 'intersect.mutect2.vcf' )), inclusion, exclusion)
intermediate_files.add(mutect2_in)
snv_mutect_out = os.sep.join(( outdir, 'snv.mutect2.vcf' ))
indel_mutect_out = os.sep.join(( outdir, 'indel.mutect2.vcf' ))
mod_mutect2.convert(mutect2_in, snv_mutect_out, indel_mutect_out)
for file_i in snv_mutect_out, indel_mutect_out:
intermediate_files.add( file_i )
snv_intermediates.append(snv_mutect_out)
indel_intermediates.append(indel_mutect_out)
intermediate_vcfs['MuTect2']['snv'] = snv_mutect_out
intermediate_vcfs['MuTect2']['indel'] = indel_mutect_out
if varscan:
import vcfModifier.modify_VarScan2 as mod_varscan2
varscan_in = bed_intersector(varscan, os.sep.join(( outdir, 'intersect.varscan.vcf' )), inclusion, exclusion)
intermediate_files.add(varscan_in)
snv_temp = os.sep.join(( outdir, 'snv.varscan.temp.vcf' ))
indel_temp = os.sep.join(( outdir, 'indel.varscan.temp.vcf' ))
snv_varscan_out = os.sep.join(( outdir, 'snv.varscan.vcf' ))
indel_varscan_out = os.sep.join(( outdir, 'indel.varscan.vcf' ))
splitVcf.split_into_snv_and_indel(varscan_in, snv_temp, indel_temp)
mod_varscan2.convert(snv_temp, snv_varscan_out)
mod_varscan2.convert(indel_temp, indel_varscan_out)
for file_i in snv_temp, indel_temp, snv_varscan_out, indel_varscan_out:
intermediate_files.add( file_i )
snv_intermediates.append(snv_varscan_out)
indel_intermediates.append(indel_varscan_out)
intermediate_vcfs['VarScan2']['snv'] = snv_varscan_out
intermediate_vcfs['VarScan2']['indel'] = indel_varscan_out
if vardict:
import vcfModifier.modify_VarDict as mod_vardict
vardict_in = bed_intersector(vardict, os.sep.join(( outdir, 'intersect.vardict.vcf' )), inclusion, exclusion)
intermediate_files.add(vardict_in)
snv_vardict_out = os.sep.join(( outdir, 'snv.vardict.vcf' ))
indel_vardict_out = os.sep.join(( outdir, 'indel.vardict.vcf'))
mod_vardict.convert(vardict_in, snv_vardict_out, indel_vardict_out)
sorted_snv_vardict_out = os.sep.join(( outdir, 'snv.sort.vardict.vcf'))
sorted_indel_vardict_out = os.sep.join(( outdir, 'indel.sort.vardict.vcf'))
vcfsorter(ref, snv_vardict_out, sorted_snv_vardict_out)
vcfsorter(ref, indel_vardict_out, sorted_indel_vardict_out)
for file_i in snv_vardict_out, indel_vardict_out, sorted_snv_vardict_out, sorted_indel_vardict_out:
intermediate_files.add( file_i )
snv_intermediates.append(sorted_snv_vardict_out)
indel_intermediates.append(sorted_indel_vardict_out)
intermediate_vcfs['VarDict']['snv'] = sorted_snv_vardict_out
intermediate_vcfs['VarDict']['indel'] = sorted_indel_vardict_out
if lofreq:
lofreq_in = bed_intersector(lofreq, os.sep.join(( outdir, 'intersect.lofreq.vcf' )), inclusion, exclusion)
intermediate_files.add(lofreq_in)
snv_lofreq_out = os.sep.join(( outdir, 'snv.lofreq.vcf' ))
indel_lofreq_out = os.sep.join(( outdir, 'indel.lofreq.vcf' ))
splitVcf.split_into_snv_and_indel(lofreq_in, snv_lofreq_out, indel_lofreq_out)
for file_i in snv_lofreq_out, indel_lofreq_out:
intermediate_files.add( file_i )
snv_intermediates.append(snv_lofreq_out)
indel_intermediates.append(indel_lofreq_out)
intermediate_vcfs['LoFreq']['snv'] = snv_lofreq_out
intermediate_vcfs['LoFreq']['indel'] = indel_lofreq_out
if scalpel:
scalpel_in = bed_intersector(scalpel, os.sep.join(( outdir, 'intersect.scalpel.vcf' )), inclusion, exclusion)
intermediate_files.add(scalpel_in)
scalpel_out = os.sep.join(( outdir, 'indel.scalpel.vcf' ))
copy_TextFile.copy(scalpel_in, scalpel_out)
intermediate_files.add(scalpel_out)
indel_intermediates.append(scalpel_out)
if strelka:
import vcfModifier.modify_ssStrelka as mod_strelka
strelka_in = bed_intersector(strelka, os.sep.join(( outdir, 'intersect.strelka.vcf' )), inclusion, exclusion)
intermediate_files.add(strelka_in)
snv_strelka_out = os.sep.join(( outdir, 'snv.strelka.vcf' ))
indel_strelka_out = os.sep.join(( outdir, 'indel.strelka.vcf' ))
mod_strelka.convert(strelka_in, snv_strelka_out, indel_strelka_out)
for file_i in snv_strelka_out, indel_strelka_out:
intermediate_files.add( file_i )
snv_intermediates.append(snv_strelka_out)
indel_intermediates.append(indel_strelka_out)
intermediate_vcfs['Strelka']['snv'] = snv_strelka_out
intermediate_vcfs['Strelka']['indel'] = indel_strelka_out
# Combine SNV/INDEL variant candidates
snv_combined = os.sep.join(( outdir, 'unsorted.CombineVariants.snv.vcf' ))
indel_combined = os.sep.join(( outdir, 'unsorted.CombineVariants.indel.vcf' ))
getUniqueVcfPositions.combine(snv_intermediates, snv_combined)
getUniqueVcfPositions.combine(indel_intermediates, indel_combined)
for file_i in snv_combined, indel_combined:
intermediate_files.add( file_i )
# Sort them:
snv_combined_sorted = os.sep.join(( outdir, 'CombineVariants.snv.vcf' ))
indel_combined_sorted = os.sep.join(( outdir, 'CombineVariants.indel.vcf' ))
vcfsorter(ref, snv_combined, snv_combined_sorted)
vcfsorter(ref, indel_combined, indel_combined_sorted)
if not keep_intermediates:
for file_i in intermediate_files:
subprocess.call( ('rm', '-v', file_i ) )
return snv_combined_sorted, indel_combined_sorted, intermediate_vcfs, intermediate_files
def combinePaired(outdir, ref, tbam, nbam, inclusion=None, exclusion=None, mutect=None, indelocator=None, mutect2=None, varscan_snv=None, varscan_indel=None, jsm=None, sniper=None, vardict=None, muse=None, lofreq_snv=None, lofreq_indel=None, scalpel=None, strelka_snv=None, strelka_indel=None, tnscope=None, platypus=None, keep_intermediates=False):
hg_dict = re.sub(r'\.fa(sta)?$', '.dict', ref)
intermediate_files = set()
snv_intermediates = []
indel_intermediates = []
intermediate_vcfs = {'MuTect2': {'snv': None, 'indel': None}, \
'VarDict': {'snv': None, 'indel': None}, \
'TNscope': {'snv': None, 'indel': None}, \
'Platypus': {'snv': None, 'indel': None} }
# Modify direct VCF outputs for merging:
if mutect or indelocator:
import vcfModifier.modify_MuTect as mod_mutect
if mutect:
mutect_in = bed_intersector(mutect, os.sep.join(( outdir, 'intersect.mutect1.vcf' )), inclusion, exclusion)
intermediate_files.add(mutect_in)
snv_mutect_out = os.sep.join(( outdir, 'snv.mutect1.vcf' ))
mod_mutect.convert(mutect_in, snv_mutect_out, tbam, nbam)
intermediate_files.add(snv_mutect_out)
snv_intermediates.append(snv_mutect_out)
if indelocator:
indelocator_in = bed_intersector(indelocator, os.sep.join(( outdir, 'intersect.indelocator.vcf' )), inclusion, exclusion)
intermediate_files.add(indelocator_in)
indel_indelocator_out = os.sep.join(( outdir, 'indel.indelocator.vcf'))
mod_mutect.convert(indelocator_in, indel_indelocator_out, tbam, nbam)
intermediate_files.add(indel_indelocator_out)
indel_intermediates.append(indel_indelocator_out)
if mutect2:
import vcfModifier.modify_MuTect2 as mod_mutect2
mutect2_in = bed_intersector(mutect2, os.sep.join(( outdir, 'intersect.mutect2.vcf' )), inclusion, exclusion)
intermediate_files.add(mutect2_in)
snv_mutect_out = os.sep.join(( outdir, 'snv.mutect2.vcf'))
indel_mutect_out = os.sep.join(( outdir, 'indel.mutect2.vcf'))
mod_mutect2.convert(mutect2_in, snv_mutect_out, indel_mutect_out, False)
for file_i in snv_mutect_out, indel_mutect_out:
intermediate_files.add( file_i )
snv_intermediates.append(snv_mutect_out)
indel_intermediates.append(indel_mutect_out)
intermediate_vcfs['MuTect2']['snv'] = snv_mutect_out
intermediate_vcfs['MuTect2']['indel'] = indel_mutect_out
if varscan_snv or varscan_indel:
import vcfModifier.modify_VarScan2 as mod_varscan2
if varscan_snv:
varscan_in = bed_intersector(varscan_snv, os.sep.join(( outdir, 'intersect.varscan.snv.vcf' )), inclusion, exclusion)
intermediate_files.add(varscan_in)
snv_varscan_out = os.sep.join(( outdir, 'snv.varscan.vcf'))
mod_varscan2.convert(varscan_in, snv_varscan_out)
intermediate_files.add(snv_varscan_out)
snv_intermediates.append(snv_varscan_out)
if varscan_indel:
varscan_in = bed_intersector(varscan_indel, os.sep.join(( outdir, 'intersect.varscan.indel.vcf' )), inclusion, exclusion)
intermediate_files.add(varscan_in)
indel_varscan_out = os.sep.join(( outdir, 'indel.varscan.vcf' ))
mod_varscan2.convert(varscan_in, indel_varscan_out)
intermediate_files.add(indel_varscan_out)
indel_intermediates.append(indel_varscan_out)
if jsm:
import vcfModifier.modify_JointSNVMix2 as mod_jsm
jsm_in = bed_intersector(jsm, os.sep.join(( outdir, 'intersect.jsm.vcf' )), inclusion, exclusion)
intermediate_files.add(jsm_in)
jsm_out = os.sep.join(( outdir, 'snv.jsm.vcf' ))
mod_jsm.convert(jsm_in, jsm_out)
intermediate_files.add(jsm_out)
snv_intermediates.append(jsm_out)
if sniper:
import vcfModifier.modify_SomaticSniper as mod_sniper
sniper_in = bed_intersector(sniper, os.sep.join(( outdir, 'intersect.sniper.vcf' )), inclusion, exclusion)
intermediate_files.add(sniper_in)
sniper_out = os.sep.join(( outdir, 'snv.somaticsniper.vcf' ))
mod_sniper.convert(sniper_in, sniper_out)
intermediate_files.add(sniper_out)
snv_intermediates.append(sniper_out)
if vardict:
import vcfModifier.modify_VarDict as mod_vardict
vardict_in = bed_intersector(vardict, os.sep.join(( outdir, 'intersect.vardict.vcf' )), inclusion, exclusion)
intermediate_files.add(vardict_in)
snv_vardict_out = os.sep.join(( outdir, 'snv.vardict.vcf' ))
indel_vardict_out = os.sep.join(( outdir, 'indel.vardict.vcf' ))
mod_vardict.convert(vardict_in, snv_vardict_out, indel_vardict_out)
sorted_snv_vardict_out = os.sep.join(( outdir, 'snv.sort.vardict.vcf' ))
sorted_indel_vardict_out = os.sep.join(( outdir, 'indel.sort.vardict.vcf' ))
vcfsorter(ref, snv_vardict_out, sorted_snv_vardict_out)
vcfsorter(ref, indel_vardict_out, sorted_indel_vardict_out)
for file_i in snv_vardict_out, indel_vardict_out, sorted_snv_vardict_out, sorted_indel_vardict_out:
intermediate_files.add(file_i)
snv_intermediates.append(sorted_snv_vardict_out)
indel_intermediates.append(sorted_indel_vardict_out)
intermediate_vcfs['VarDict']['snv'] = sorted_snv_vardict_out
intermediate_vcfs['VarDict']['indel'] = sorted_indel_vardict_out
if muse:
muse_in = bed_intersector(muse, os.sep.join(( outdir, 'intersect.muse.vcf' )), inclusion, exclusion)
intermediate_files.add(muse_in)
muse_out = os.sep.join(( outdir, 'snv.muse.vcf' ))
copy_TextFile.copy(muse_in, muse_out)
intermediate_files.add(muse_out)
snv_intermediates.append(muse_out)
if lofreq_snv:
lofreq_in = bed_intersector(lofreq_snv, os.sep.join(( outdir, 'intersect.lofreq.snv.vcf' )), inclusion, exclusion)
intermediate_files.add(lofreq_in)
snv_lofreq_out = os.sep.join(( outdir, 'snv.lofreq.vcf' ))
copy_TextFile.copy(lofreq_in, snv_lofreq_out)
intermediate_files.add(snv_lofreq_out)
snv_intermediates.append(snv_lofreq_out)
if lofreq_indel:
lofreq_in = bed_intersector(lofreq_indel, os.sep.join(( outdir, 'intersect.lofreq.indel.vcf' )), inclusion, exclusion)
intermediate_files.add(lofreq_in)
indel_lofreq_out = os.sep.join(( outdir, 'indel.lofreq.vcf' ))
copy_TextFile.copy(lofreq_in, indel_lofreq_out)
intermediate_files.add(indel_lofreq_out)
indel_intermediates.append(indel_lofreq_out)
if scalpel:
scalpel_in = bed_intersector(scalpel, os.sep.join(( outdir, 'intersect.scalpel.vcf' )), inclusion, exclusion)
intermediate_files.add(scalpel_in)
scalpel_out = os.sep.join(( outdir, 'indel.scalpel.vcf' ))
copy_TextFile.copy(scalpel_in, scalpel_out)
intermediate_files.add(scalpel_out)
indel_intermediates.append(scalpel_out)
if strelka_snv or strelka_indel:
import vcfModifier.modify_Strelka as mod_strelka
if strelka_snv:
strelka_in = bed_intersector(strelka_snv, os.sep.join(( outdir, 'intersect.strelka.snv.vcf' )), inclusion, exclusion)
intermediate_files.add(strelka_in)
snv_strelka_out = os.sep.join(( outdir, 'snv.strelka.vcf' ))
mod_strelka.convert(strelka_in, snv_strelka_out)
intermediate_files.add(snv_strelka_out)
snv_intermediates.append(snv_strelka_out)
if strelka_indel:
strelka_in = bed_intersector(strelka_indel, os.sep.join(( outdir, 'intersect.strelka.indel.vcf' )), inclusion, exclusion)
intermediate_files.add(strelka_in)
indel_strelka_out = os.sep.join(( outdir, 'indel.strelka.vcf' ))
mod_strelka.convert(strelka_in, indel_strelka_out)
intermediate_files.add(indel_strelka_out)
indel_intermediates.append(indel_strelka_out)
if tnscope:
import vcfModifier.modify_MuTect2 as mod_mutect2
tnscope_in = bed_intersector(tnscope, os.sep.join(( outdir, 'intersect.tnscope.vcf' )), inclusion, exclusion)
intermediate_files.add(tnscope_in)
snv_tnscope_out = os.sep.join(( outdir, 'snv.tnscope.vcf' ))
indel_tnscope_out = os.sep.join(( outdir, 'indel.tnscope.vcf' ))
mod_mutect2.convert(tnscope_in, snv_tnscope_out, indel_tnscope_out, True)
for file_i in snv_tnscope_out, indel_tnscope_out:
intermediate_files.add( file_i )
snv_intermediates.append(snv_tnscope_out)
indel_intermediates.append(indel_tnscope_out)
intermediate_vcfs['TNscope']['snv'] = snv_tnscope_out
intermediate_vcfs['TNscope']['indel'] = indel_tnscope_out
if platypus:
platypus_in = bed_intersector(platypus, os.sep.join(( outdir, 'intersect.platypus.vcf' )), inclusion, exclusion)
intermediate_files.add(platypus_in)
snv_platypus_out = os.sep.join(( outdir, 'snv.platypus.vcf' ))
indel_platypus_out = os.sep.join(( outdir, 'indel.platypus.vcf' ))
splitVcf.split_into_snv_and_indel(platypus_in, snv_platypus_out, indel_platypus_out)
for file_i in snv_platypus_out, indel_platypus_out:
intermediate_files.add( file_i )
snv_intermediates.append(snv_platypus_out)
indel_intermediates.append(indel_platypus_out)
intermediate_vcfs['Platypus']['snv'] = snv_platypus_out
intermediate_vcfs['Platypus']['indel'] = indel_platypus_out
# Combine SNV/INDEL variant candidates
snv_combined = os.sep.join(( outdir, 'unsorted.CombineVariants.snv.vcf' ))
indel_combined = os.sep.join(( outdir, 'unsorted.CombineVariants.indel.vcf' ))
getUniqueVcfPositions.combine(snv_intermediates, snv_combined)
getUniqueVcfPositions.combine(indel_intermediates, indel_combined)
for file_i in snv_combined, indel_combined:
intermediate_files.add( file_i )
# Sort them:
snv_combined_sorted = os.sep.join(( outdir, 'CombineVariants.snv.vcf' ))
indel_combined_sorted = os.sep.join(( outdir, 'CombineVariants.indel.vcf' ))
vcfsorter(ref, snv_combined, snv_combined_sorted)
vcfsorter(ref, indel_combined, indel_combined_sorted)
if not keep_intermediates:
for file_i in intermediate_files:
subprocess.call( ('rm', '-v', file_i ) )
return snv_combined_sorted, indel_combined_sorted, intermediate_vcfs, intermediate_files
def combinePaired(outdir,
ref,
tbam,
nbam,
inclusion=None,
exclusion=None,
mutect=None,
indelocator=None,
mutect2=None,
varscan_snv=None,
varscan_indel=None,
jsm=None,
sniper=None,
vardict=None,
muse=None,
lofreq_snv=None,
lofreq_indel=None,
scalpel=None,
strelka_snv=None,
strelka_indel=None,
tnscope=None,
keep_intermediates=False):
hg_dict = re.sub(r'\.fa(sta)?$', '.dict', ref)
intermediate_files = set()
snv_intermediates = []
indel_intermediates = []
intermediate_vcfs = {'MuTect2':{'snv': None, 'indel': None}, \
'VarDict':{'snv': None, 'indel': None}, \
'TNscope':{'snv': None, 'indel': None}, }
# Modify direct VCF outputs for merging:
if mutect or indelocator:
import vcfModifier.modify_MuTect as mod_mutect
if mutect:
if exclusion:
mutect_ex = bed_exclude(
mutect, exclusion,
os.sep.join((outdir, 'snv.mutect1.ex.vcf')))
intermediate_files.add(mutect_ex)
else:
mutect_ex = mutect
if inclusion:
mutect_in = bed_include(
mutect_ex, inclusion,
os.sep.join((outdir, 'snv.mutect1.in.vcf')))
intermediate_files.add(mutect_in)
else:
mutect_in = mutect_ex
snv_mutect_out = os.sep.join((outdir, 'snv.mutect1.vcf'))
mod_mutect.convert(mutect_in, snv_mutect_out, tbam, nbam)
intermediate_files.add(snv_mutect_out)
snv_intermediates.append(snv_mutect_out)
if indelocator:
if exclusion:
indelocator_ex = bed_exclude(
indelocator, exclusion,
os.sep.join((outdir, 'indel.indelocator.ex.vcf')))
intermediate_files.add(indelocator_ex)
else:
indelocator_ex = indelocator
if inclusion:
indelocator_in = bed_include(
indelocator_ex, inclusion,
os.sep.join((outdir, 'indel.indelocator.in.vcf')))
intermediate_files.add(indelocator_in)
else:
indelocator_in = indelocator_ex
indel_indelocator_out = os.sep.join(
(outdir, 'indel.indelocator.vcf'))
mod_mutect.convert(indelocator_in, indel_indelocator_out, tbam,
nbam)
intermediate_files.add(indel_indelocator_out)
indel_intermediates.append(indel_indelocator_out)
if mutect2:
import vcfModifier.modify_MuTect2 as mod_mutect2
if exclusion:
mutect2_ex = bed_exclude(mutect2, exclusion,
os.sep.join((outdir, 'mutect.ex.vcf')))
intermediate_files.add(mutect2_ex)
else:
mutect2_ex = mutect2
if inclusion:
mutect2_in = bed_include(mutect2_ex, inclusion,
os.sep.join((outdir, 'mutect.in.vcf')))
intermediate_files.add(mutect2_in)
else:
mutect2_in = mutect2_ex
snv_mutect_out = os.sep.join((outdir, 'snv.mutect.vcf'))
indel_mutect_out = os.sep.join((outdir, 'indel.mutect.vcf'))
mod_mutect2.convert(mutect2_in, snv_mutect_out, indel_mutect_out,
False)
for file_i in snv_mutect_out, indel_mutect_out:
intermediate_files.add(file_i)
snv_intermediates.append(snv_mutect_out)
indel_intermediates.append(indel_mutect_out)
intermediate_vcfs['MuTect2']['snv'] = snv_mutect_out
intermediate_vcfs['MuTect2']['indel'] = indel_mutect_out
if varscan_snv or varscan_indel:
import vcfModifier.modify_VarScan2 as mod_varscan2
if varscan_snv:
if exclusion:
varscan_ex = bed_exclude(
varscan_snv, exclusion,
os.sep.join((outdir, 'snv.varscan.ex.vcf')))
intermediate_files.add(varscan_ex)
else:
varscan_ex = varscan_snv
if inclusion:
varscan_in = bed_include(
varscan_ex, inclusion,
os.sep.join((outdir, 'snv.varscan.in.vcf')))
intermediate_files.add(varscan_in)
else:
varscan_in = varscan_ex
snv_varscan_out = os.sep.join((outdir, 'snv.varscan.vcf'))
mod_varscan2.convert(varscan_in, snv_varscan_out)
intermediate_files.add(snv_varscan_out)
snv_intermediates.append(snv_varscan_out)
if varscan_indel:
if exclusion:
varscan_ex = bed_exclude(
varscan_indel, exclusion,
os.sep.join((outdir, 'indel.varscan.ex.vcf')))
intermediate_files.add(varscan_ex)
else:
varscan_ex = varscan_indel
if inclusion:
varscan_in = bed_include(
varscan_ex, inclusion,
os.sep.join((outdir, 'indel.varscan.in.vcf')))
intermediate_files.add(varscan_in)
else:
varscan_in = varscan_ex
indel_varscan_out = os.sep.join((outdir, 'indel.varscan.vcf'))
mod_varscan2.convert(varscan_in, indel_varscan_out)
intermediate_files.add(indel_varscan_out)
indel_intermediates.append(indel_varscan_out)
if jsm:
import vcfModifier.modify_JointSNVMix2 as mod_jsm
if exclusion:
jsm_ex = bed_exclude(jsm, exclusion,
os.sep.join((outdir, 'snv.jsm.ex.vcf')))
intermediate_files.add(jsm_ex)
else:
jsm_ex = jsm
if inclusion:
jsm_in = bed_include(jsm_ex, inclusion,
os.sep.join((outdir, 'snv.jsm.in.vcf')))
intermediate_files.add(jsm_in)
else:
jsm_in = jsm_ex
jsm_out = os.sep.join((outdir, 'snv.jsm.vcf'))
mod_jsm.convert(jsm_in, jsm_out)
intermediate_files.add(jsm_out)
snv_intermediates.append(jsm_out)
if sniper:
import vcfModifier.modify_SomaticSniper as mod_sniper
if exclusion:
sniper_ex = bed_exclude(
sniper, exclusion,
os.sep.join((outdir, 'snv.somaticsniper.ex.vcf')))
intermediate_files.add(sniper_ex)
else:
sniper_ex = sniper
if inclusion:
sniper_in = bed_include(
sniper_ex, inclusion,
os.sep.join((outdir, 'snv.somaticsniper.in.vcf')))
intermediate_files.add(sniper_in)
else:
sniper_in = sniper_ex
sniper_out = os.sep.join((outdir, 'snv.somaticsniper.vcf'))
mod_sniper.convert(sniper_in, sniper_out)
intermediate_files.add(sniper_out)
snv_intermediates.append(sniper_out)
if vardict:
import vcfModifier.modify_VarDict as mod_vardict
if exclusion:
vardict_ex = bed_exclude(vardict, exclusion,
os.sep.join((outdir, 'vardict.ex.vcf')))
intermediate_files.add(vardict_ex)
else:
vardict_ex = vardict
if inclusion:
vardict_in = bed_include(vardict_ex, inclusion,
os.sep.join((outdir, 'vardict.in.vcf')))
intermediate_files.add(vardict_in)
else:
vardict_in = vardict_ex
snv_vardict_out = os.sep.join((outdir, 'snv.vardict.vcf'))
indel_vardict_out = os.sep.join((outdir, 'indel.vardict.vcf'))
mod_vardict.convert(vardict_in, snv_vardict_out, indel_vardict_out)
sorted_snv_vardict_out = os.sep.join((outdir, 'snv.sort.vardict.vcf'))
sorted_indel_vardict_out = os.sep.join(
(outdir, 'indel.sort.vardict.vcf'))
vcfsorter(ref, snv_vardict_out, sorted_snv_vardict_out)
vcfsorter(ref, indel_vardict_out, sorted_indel_vardict_out)
for file_i in snv_vardict_out, indel_vardict_out, sorted_snv_vardict_out, sorted_indel_vardict_out:
intermediate_files.add(file_i)
snv_intermediates.append(sorted_snv_vardict_out)
indel_intermediates.append(sorted_indel_vardict_out)
intermediate_vcfs['VarDict']['snv'] = sorted_snv_vardict_out
intermediate_vcfs['VarDict']['indel'] = sorted_indel_vardict_out
if muse:
if exclusion:
muse_ex = bed_exclude(muse, exclusion,
os.sep.join((outdir, 'snv.muse.ex.vcf')))
intermediate_files.add(muse_ex)
else:
muse_ex = muse
if inclusion:
muse_in = bed_include(muse_ex, inclusion,
os.sep.join((outdir, 'snv.muse.in.vcf')))
intermediate_files.add(muse_in)
else:
muse_in = muse_ex
muse_out = os.sep.join((outdir, 'snv.muse.vcf'))
copy_TextFile.copy(muse_in, muse_out)
intermediate_files.add(muse_out)
snv_intermediates.append(muse_out)
if lofreq_snv:
if exclusion:
lofreq_ex = bed_exclude(lofreq_snv, exclusion,
os.sep.join((outdir, 'snv.lofreq.ex.vcf')))
intermediate_files.add(lofreq_ex)
else:
lofreq_ex = lofreq_snv
if inclusion:
lofreq_in = bed_include(lofreq_ex, inclusion,
os.sep.join((outdir, 'snv.lofreq.in.vcf')))
intermediate_files.add(lofreq_in)
else:
lofreq_in = lofreq_ex
snv_lofreq_out = os.sep.join((outdir, 'snv.lofreq.vcf'))
copy_TextFile.copy(lofreq_in, snv_lofreq_out)
intermediate_files.add(snv_lofreq_out)
snv_intermediates.append(snv_lofreq_out)
if lofreq_indel:
if exclusion:
lofreq_ex = bed_exclude(
lofreq_indel, exclusion,
os.sep.join((outdir, 'indel.lofreq.ex.vcf')))
intermediate_files.add(lofreq_ex)
else:
lofreq_ex = lofreq_snv
if inclusion:
lofreq_in = bed_include(
lofreq_ex, inclusion,
os.sep.join((outdir, 'indel.lofreq.in.vcf')))
intermediate_files.add(lofreq_in)
else:
lofreq_in = lofreq_ex
indel_lofreq_out = os.sep.join((outdir, 'indel.lofreq.vcf'))
copy_TextFile.copy(lofreq_in, indel_lofreq_out)
intermediate_files.add(indel_lofreq_out)
indel_intermediates.append(indel_lofreq_out)
if scalpel:
if exclusion:
scalpel_ex = bed_exclude(
scalpel, exclusion,
os.sep.join((outdir, 'indel.scalpel.ex.vcf')))
intermediate_files.add(scalpel_ex)
else:
scalpel_ex = scalpel
if inclusion:
scalpel_in = bed_include(
scalpel_ex, inclusion,
os.sep.join((outdir, 'indel.scalpel.in.vcf')))
intermediate_files.add(scalpel_in)
else:
scalpel_in = scalpel_ex
scalpel_out = os.sep.join((outdir, 'indel.scalpel.vcf'))
copy_TextFile.copy(scalpel_in, scalpel_out)
intermediate_files.add(scalpel_out)
indel_intermediates.append(scalpel_out)
if strelka_snv or strelka_indel:
import vcfModifier.modify_Strelka as mod_strelka
if strelka_snv:
if exclusion:
strelka_ex = bed_exclude(
strelka_snv, exclusion,
os.sep.join((outdir, 'snv.strelka.ex.vcf')))
intermediate_files.add(strelka_ex)
else:
strelka_ex = strelka_snv
if inclusion:
strelka_in = bed_include(
strelka_ex, inclusion,
os.sep.join((outdir, 'snv.strelka.in.vcf')))
intermediate_files.add(strelka_in)
else:
strelka_in = strelka_ex
snv_strelka_out = os.sep.join((outdir, 'snv.strelka.vcf'))
mod_strelka.convert(strelka_in, snv_strelka_out)
intermediate_files.add(snv_strelka_out)
snv_intermediates.append(snv_strelka_out)
if strelka_indel:
if exclusion:
strelka_ex = bed_exclude(
strelka_indel, exclusion,
os.sep.join((outdir, 'indel.strelka.ex.vcf')))
intermediate_files.add(strelka_ex)
else:
strelka_ex = strelka_snv
if inclusion:
strelka_in = bed_include(
strelka_ex, inclusion,
os.sep.join((outdir, 'indel.strelka.in.vcf')))
intermediate_files.add(strelka_in)
else:
strelka_in = strelka_ex
indel_strelka_out = os.sep.join((outdir, 'indel.strelka.vcf'))
mod_strelka.convert(strelka_in, indel_strelka_out)
intermediate_files.add(indel_strelka_out)
indel_intermediates.append(indel_strelka_out)
if tnscope:
import vcfModifier.modify_MuTect2 as mod_mutect2
if exclusion:
tnscope_ex = bed_exclude(tnscope, exclusion,
os.sep.join((outdir, 'tnscope.ex.vcf')))
intermediate_files.add(tnscope_ex)
else:
tnscope_ex = tnscope
if inclusion:
tnscope_in = bed_include(tnscope_ex, inclusion,
os.sep.join((outdir, 'tnscope.in.vcf')))
intermediate_files.add(tnscope_in)
else:
tnscope_in = tnscope_ex
snv_tnscope_out = os.sep.join((outdir, 'snv.tnscope.vcf'))
indel_tnscope_out = os.sep.join((outdir, 'indel.tnscope.vcf'))
mod_mutect2.convert(tnscope_in, snv_tnscope_out, indel_tnscope_out,
True)
for file_i in snv_tnscope_out, indel_tnscope_out:
intermediate_files.add(file_i)
snv_intermediates.append(snv_tnscope_out)
indel_intermediates.append(indel_tnscope_out)
intermediate_vcfs['TNscope']['snv'] = snv_tnscope_out
intermediate_vcfs['TNscope']['indel'] = indel_tnscope_out
# Combine SNV/INDEL variant candidates
snv_combined = os.sep.join((outdir, 'unsorted.CombineVariants.snv.vcf'))
indel_combined = os.sep.join(
(outdir, 'unsorted.CombineVariants.indel.vcf'))
getUniqueVcfPositions.combine(snv_intermediates, snv_combined)
getUniqueVcfPositions.combine(indel_intermediates, indel_combined)
for file_i in snv_combined, indel_combined:
intermediate_files.add(file_i)
# Sort them:
snv_combined_sorted = os.sep.join((outdir, 'CombineVariants.snv.vcf'))
indel_combined_sorted = os.sep.join((outdir, 'CombineVariants.indel.vcf'))
vcfsorter(ref, snv_combined, snv_combined_sorted)
vcfsorter(ref, indel_combined, indel_combined_sorted)
if not keep_intermediates:
for file_i in intermediate_files:
subprocess.call(('rm', '-v', file_i))
return snv_combined_sorted, indel_combined_sorted, intermediate_vcfs, intermediate_files