def pileup_depth(self, pileup_depth):
md5_file = '%s.md5' % (pileup_depth)
if md5sum_check(pileup_depth, md5_file):
log_progress(__modname__,
'Get Pileup Depth already finished!!!',
f=self._log_file)
log_version(__modname__,
self._sw['samtools_ver'],
f=self._log_file)
else:
log_progress(__modname__,
'Get Pileup Depth start',
f=self._log_file)
log_version(__modname__,
self._sw['samtools_ver'],
f=self._log_file)
exec_cmd = [
self._sw['samtools'], 'depth', '-a', '-q', '0', '-Q', '1',
'-d', '1000000', '-b', self._target_bed, '--reference',
self._sw['hg19'], self._final_bam
]
run_command_file_handle(__modname__, exec_cmd, self._log_file, 'w',
pileup_depth)
run_command_md5sum(__modname__, self._log_file, pileup_depth,
md5_file)
log_progress(__modname__,
'Get Pileup Depth finished',
f=self._log_file)
def vcf_post_processing(self, input_file, refined_vcf):
refined_vcf_md5 = '{0}.md5'.format(refined_vcf)
if md5sum_check(refined_vcf, refined_vcf_md5):
log_progress(__modname__,
'VCF post processing already finished!!!',
f=self._log_file)
log_version(__modname__, self._sw['vt_ver'], f=self._log_file)
else:
log_progress(__modname__,
'VCF post processing start',
f=self._log_file)
log_version(__modname__, self._sw['vt_ver'], f=self._log_file)
if os.path.exists(refined_vcf):
os.remove(refined_vcf)
exec_cmd = [
'{0} normalize -r {1} {2}'.format(self._sw['vt'],
self._sw['hg19'],
input_file),
'{0} decompose -s -'.format(self._sw['vt']),
]
run_command_pipe_file_handle(__modname__, exec_cmd, self._log_file,
'w', refined_vcf)
run_command_md5sum(__modname__, self._log_file, refined_vcf,
refined_vcf_md5)
log_progress(__modname__,
'VCF post processing finished',
f=self._log_file)
def remove_reference_info(self, input_file, remove_ref_vcf):
remove_ref_vcf_md5 = '{0}.md5'.format(remove_ref_vcf)
# remove only reference...
if md5sum_check(remove_ref_vcf, remove_ref_vcf_md5):
log_progress(__modname__,
'Remove only reference in VCF already finished!!!',
f=self._log_file)
log_version(__modname__,
self._sw['vcftools_ver'],
f=self._log_file)
log_version(__modname__, self._sw['vt_ver'], f=self._log_file)
else:
log_progress(__modname__,
'Remove only reference in VCF start',
f=self._log_file)
log_version(__modname__,
self._sw['vcftools_ver'],
f=self._log_file)
log_version(__modname__, self._sw['vt_ver'], f=self._log_file)
if os.path.exists(remove_ref_vcf):
os.remove(remove_ref_vcf)
exec_cmd = [
'{0} --vcf {1} --recode --stdout'.format(
self._sw['vcftools'], input_file), 'grep -v "0[/|]0"',
'grep -v "\.[/|]\."', '{0} sort -'.format(self._sw['vt'])
]
run_command_pipe_file_handle(__modname__, exec_cmd, self._log_file,
'w', remove_ref_vcf)
run_command_md5sum(__modname__, self._log_file, remove_ref_vcf,
remove_ref_vcf_md5)
log_progress(__modname__,
'Remove only reference in VCF finished',
f=self._log_file)
def run_snpEff(self, input_file, output_file):
snpeff_tmp_out = join(self._variant_dir,
'{0}_snpeff_tmp.vcf'.format(self._sample_name))
output_md5 = '{0}.md5'.format(output_file)
if md5sum_check(output_file, output_md5):
log_progress(__modname__,
'snpEff gene annotation already finished!!!',
f=self._log_file)
log_version(__modname__, self._sw['snpeff_ver'], f=self._log_file)
else:
log_progress(__modname__,
'snpEff gene annotation start',
f=self._log_file)
log_version(__modname__, self._sw['snpeff_ver'], f=self._log_file)
if os.path.exists(snpeff_tmp_out):
os.remove(snpeff_tmp_out)
if os.path.exists(output_file):
os.remove(output_file)
exec_cmd1 = [
self._sw['java'], '-Xmx4g', '-XX:ParallelGCThreads={0}'.format(
self._pe_core), '-Djava.io.tmpdir={0}'.format(
self._sample_tmp_dir), '-jar', self._sw['snpeff'],
'ann', 'hg19ngb', '-no-downstream', '-no-upstream', '-noStats',
'-no', 'INTERGENIC', '-no', 'INTERGENIC_CONSERVED', '-no',
'INTRAGENIC', '-no', 'RARE_AMINO_ACID', '-no', 'TRANSCRIPT',
'-no', 'TRANSCRIPT_DELETED', '-no', 'REGULATION', '-no',
'NEXT_PROT', '-no', 'PROTEIN_STRUCTURAL_INTERACTION_LOCUS',
'-no', 'PROTEIN_PROTEIN_INTERACTION_LOCUS', input_file
]
run_command_file_handle(__modname__, exec_cmd1, self._log_file,
'w', snpeff_tmp_out)
exec_cmd2 = [
self._sw['bcftools'], 'view', '-i', 'INFO/ANN!="."',
snpeff_tmp_out
]
run_command_file_handle(__modname__, exec_cmd2, self._log_file,
'w', output_file)
run_command_md5sum(__modname__, self._log_file, output_file,
output_md5)
log_progress(__modname__,
'snpEff gene annotation finished',
f=self._log_file)
def run_dbnsfp_annotation(self, input_file, output_file):
dbnsfp_tmp_vcf = join(self._variant_dir,
'{0}_dbnsfp_tmp.vcf'.format(self._sample_name))
output_md5 = '{0}.md5'.format(output_file)
if md5sum_check(output_file, output_md5):
log_progress(__modname__,
'dbNSFP annotation already finished!!!',
f=self._log_file)
log_version(__modname__,
self._sw['dbnsfp_db_ver'],
f=self._log_file)
else:
log_progress(__modname__,
'dbNSFP annotation start',
f=self._log_file)
log_version(__modname__,
self._sw['dbnsfp_db_ver'],
f=self._log_file)
exec_cmd1 = [
self._sw['java'], '-Xmx4g',
'-XX:ParallelGCThreads={0}'.format(self._pe_core),
'-Djava.io.tmpdir={0}'.format(self._sample_tmp_dir), '-jar',
self._sw['snpsift'], 'dbnsfp', '-f',
'aapos,aapos_SIFT,aapos_FATHMM,Uniprot_acc,Interpro_domain,SIFT_pred,SIFT_score,LRT_pred,MutationTaster_pred,MutationTaster_score,GERP++_NR,GERP++_RS,phastCons100way_vertebrate,MutationAssessor_pred,FATHMM_pred,PROVEAN_pred,MetaSVM_pred,Polyphen2_HDIV_pred,Polyphen2_HDIV_score,Polyphen2_HVAR_pred,Polyphen2_HVAR_score,CADD_phred',
'-db', self._sw['dbnsfp_db'], input_file
]
run_command_file_handle(__modname__, exec_cmd1, self._log_file,
'w', dbnsfp_tmp_vcf)
exec_cmd2 = [
'python', self._sw['ngb_transcript_dbNSFP'], '-o', output_file,
dbnsfp_tmp_vcf
]
run_command(__modname__, exec_cmd2, self._log_file)
run_command_md5sum(__modname__, self._log_file, output_file,
output_md5)
run_command(__modname__,
['rm', '-rf', '{0}/*'.format(self._sample_tmp_dir)],
self._log_file)
log_progress(__modname__,
'dbNSFP annotation finished',
f=self._log_file)
def generate_tdf_file(self, final_bam):
tdf_file = '{0}.tdf'.format(final_bam)
tdf_file_md5 = '{0}.md5'.format(tdf_file)
if md5sum_check(tdf_file, tdf_file_md5):
log_progress(__modname__, 'TDF file generation already finished!!!', f=self._log_file)
log_version(__modname__, self._sw['igvtools_ver'], f=self._log_file)
else:
log_progress(__modname__, 'TDF file generation start', f=self._log_file)
log_version(__modname__, self._sw['igvtools_ver'], f=self._log_file)
if os.path.exists(tdf_file):
os.remove(tdf_file)
exec_cmd = [
self._sw['igvtools'],
'count',
final_bam,
tdf_file,
'hg19'
]
run_command(__modname__, exec_cmd, self._log_file)
run_command_md5sum(__modname__, self._log_file, tdf_file, tdf_file_md5)
log_progress(__modname__, 'TDF file generation finished', f=self._log_file)
def run_TST170_pipeline(self, TST170_completed_file_path):
with open(self._status_log_file, 'w') as f:
f.write('[STATUS] TruSightTumor170 pipeline\n[PROGRESS] 5')
log_progress(__modname__, 'Run the TruSightTumor170 pipeline', f=self._log_file)
if os.path.exists(TST170_completed_file_path):
log_progress(__modname__, 'TruSightTumor170 pipeline already finished', f=self._log_file)
else:
log_progress(__modname__, 'TruSightTumor170 pipeline start', f=self._log_file)
exec_cmd = [
'docker', 'run',
'-t',
'--rm',
'--name', self._run_id,
'-v', '/etc/localtime:/etc/localtime',
'-v', '{0}:/data'.format(self._run_dir),
'-v', '{0}:/genomes'.format(self._TST170_Resources),
'-v', '{0}:/analysis'.format(self._output_dir),
'ngb_tst170:1.0.0.0'
]
run_command(__modname__, exec_cmd, self._log_file)
sleep(10)
# check TST170 output directory
log_version(__modname__, 'software,TST170 Local App,v1.0,Oct.2017,http://sapac.support.illumina.com/downloads/trusight-tumor-170-local-app-documentation.html?langsel=/kr/,None', f=self._log_file)
_dirs = os.listdir(self._output_dir)
tst170_dir = None
for _dir in _dirs:
if _dir.startswith("TruSightTumor170_Analysis_"):
tst170_dir = os.path.join(self._output_dir, _dir)
if tst170_dir is not None and os.path.exists(join(tst170_dir, 'Summary.tsv')):
with open(TST170_completed_file_path, 'w') as completed_file:
completed_file.write('TruSightTumor170 pipeline finished')
log_progress(__modname__, 'TruSightTumor170 pipeline finished', f=self._log_file)
return tst170_dir
def run_clinvar_annotation(self, input_file, output_file):
clinvar_tmp_vcf = join(self._variant_dir,
'{0}_clinvar_tmp.vcf'.format(self._sample_name))
output_md5 = '{0}.md5'.format(output_file)
if md5sum_check(output_file, output_md5):
log_progress(__modname__,
'ClinVar annotation already finished!!!',
f=self._log_file)
log_version(__modname__,
self._sw['ngb_clinvar_ver'],
f=self._log_file)
else:
log_progress(__modname__,
'ClinVar annotation start',
f=self._log_file)
log_version(__modname__,
self._sw['ngb_clinvar_ver'],
f=self._log_file)
exec_cmd1 = [
'python', self._sw['ngb_anno_clinvar'], '--dbfile',
self._sw['ngb_clinvar_db'], '--infoVCF',
self._sw['clinvar_compact_header'], '--inVCF', input_file,
'--outVCF', clinvar_tmp_vcf
]
run_command(__modname__, exec_cmd1, self._log_file)
exec_cmd2 = [
'python', self._sw['ngb_clinvar_variation'], '-d',
self._sw['ngb_clinvar_ref'], '-o', output_file, clinvar_tmp_vcf
]
run_command(__modname__, exec_cmd2, self._log_file)
run_command_md5sum(__modname__, self._log_file, output_file,
output_md5)
log_progress(__modname__,
'ClinVar annotation finished',
f=self._log_file)
def run_annotation(self, db_name, target_vcf, target_vcf_ver, _info, _name,
input_file, output_file):
output_md5 = '{0}.md5'.format(output_file)
if md5sum_check(output_file, output_md5):
log_progress(__modname__,
'{0} annotation already finished!!!'.format(db_name),
f=self._log_file)
log_version(__modname__, target_vcf_ver, f=self._log_file)
else:
log_progress(__modname__,
'{0} annotation start'.format(db_name),
f=self._log_file)
log_version(__modname__, target_vcf_ver, f=self._log_file)
exec_cmd = [
self._sw['java'], '-Xmx4g',
'-XX:ParallelGCThreads={0}'.format(self._pe_core),
'-Djava.io.tmpdir={0}'.format(self._sample_tmp_dir), '-jar',
self._sw['snpsift'], 'annotate'
]
if _info != '':
exec_cmd.append('-info')
exec_cmd.append(_info)
if _name != '':
exec_cmd.append('-name')
exec_cmd.append(_name)
exec_cmd.append(target_vcf)
exec_cmd.append(input_file)
run_command_file_handle(__modname__, exec_cmd, self._log_file, 'w',
output_file)
run_command_md5sum(__modname__, self._log_file, output_file,
output_md5)
run_command(__modname__,
['rm', '-rf', '{0}/*'.format(self._sample_tmp_dir)],
self._log_file)
log_progress(__modname__,
'{0} annotation finished'.format(db_name),
f=self._log_file)