def run_base_recalibrator(bam, known_sites, ref_fn, recal_table, log_file):
'''
Run GATK BaseRecalibrator.
'''
quiet_call([
'java',
'-Xmx8g',
'-jar',
PATHS['gatk'],
'-T',
'BaseRecalibrator',
'-I',
bam,
'-knownSites',
known_sites,
'--out',
recal_table,
'-R',
ref_fn,
'-cov',
'RepeatLengthCovariate',
'-cov',
'RepeatUnitCovariate',
'-log',
log_file,
])
def create_sequence_dictionary(ref_fn):
'''
Run Picard CreateSequenceDictionary if none found.
'''
if not os.path.exists(ref_fn.split('.fa')[0] + '.dict'):
quiet_call([
PATHS['java'], '-jar',
PATHS['picard'],
'CreateSequenceDictionary',
'R=' + ref_fn,
'O=' + ref_fn.split('.fa')[0] + '.dict',
])
def realigner_target_creator(ref_fn, in_bam, intervals):
'''
Run GATK RealignerTargetCreator.
'''
quiet_call([
PATHS['java'], '-jar',
PATHS['gatk'],
'-R', ref_fn,
'-T', 'RealignerTargetCreator',
'-I', in_bam,
'-o', intervals,
])
def align(fastq_1, ref_fn, output_fn, fastq_2=None, p=1):
'''
Align reads using Bowtie2.
'''
assert os.path.exists(fastq_1)
if fastq_2:
assert os.path.exists(fastq_2)
if fastq_2:
quiet_call([
PATHS['bowtie2'],
'-q',
'--phred33',
'-p',
str(p),
'-I',
'0',
'-X',
'1000',
'--fr',
'--local',
'--sensitive-local',
'-S',
output_fn,
'-x',
ref_fn,
'-1',
fastq_1,
'-2',
fastq_2,
])
else:
quiet_call([
PATHS['bowtie2'],
'-q',
'--phred33',
'-p',
str(p),
'-I',
'0',
'-X',
'1000',
'--local',
'--sensitive-local',
'-S',
output_fn,
'-x',
ref_fn,
'-U',
fastq_1,
])
def run_print_reads_bqsr(input_bam, ref_fn, recal_table, output_bam, log_file):
'''
Run GATK PrintReads, observing BQSR relcaibration table.
'''
quiet_call([
PATHS['java'], '-jar',
PATHS['gatk'],
'-T', 'PrintReads',
'-I', input_bam,
'-R', ref_fn,
'-BQSR', recal_table,
'-o', output_bam,
'-log', log_file,
])
def fix_mate_information(in_bam, out_bam, tmp_dir, max_records):
'''
Run Picard FixMateInformation.
'''
quiet_call([
PATHS['java'], '-jar',
PATHS['picard'],
'FixMateInformation',
'VALIDATION_STRINGENCY=SILENT',
'SO=coordinate',
'I=' + in_bam,
'O=' + out_bam,
'TMP_DIR=' + tmp_dir,
'MAX_RECORDS_IN_RAM=' + str(max_records),
])
def fix_mate_information(in_bam, out_bam):
'''
Run Picard FixMateInformation.
'''
quiet_call([
'java',
'-Xmx2g',
'-jar',
PATHS['picard'],
'FixMateInformation',
'VALIDATION_STRINGENCY=SILENT',
'SO=coordinate',
'I=' + in_bam,
'O=' + out_bam,
])
def indel_realigner(ref_fn, log, in_bam, intervals, realigned_bam):
'''
Run GATK IndelRealigner.
'''
quiet_call([
PATHS['java'], '-jar',
PATHS['gatk'],
'-R', ref_fn,
'-T', 'IndelRealigner',
'--maxReadsForRealignment', '100000',
'-log', log,
'-I', in_bam,
'-targetIntervals', intervals,
'-o', realigned_bam,
])
def build(ref_fn):
'''
Build Bowtie2 index for reference if none found.
'''
if not (os.path.isfile(ref_fn + '.1.bt2')
and os.path.isfile(ref_fn + '.2.bt2')
and os.path.isfile(ref_fn + '.3.bt2')
and os.path.isfile(ref_fn + '.4.bt2')
and os.path.isfile(ref_fn + '.rev.1.bt2')
and os.path.isfile(ref_fn + '.rev.2.bt2')):
quiet_call([
PATHS['bowtie2_build'],
ref_fn,
ref_fn,
])
def deduplicate(in_bam, out_bam, metrics_file, tmp_dir, max_records):
'''
Run Picard MarkDuplicates.
'''
quiet_call([
PATHS['java'], '-jar',
PATHS['picard'],
'MarkDuplicates',
'VALIDATION_STRINGENCY=SILENT',
'REMOVE_DUPLICATES=TRUE',
'I=' + in_bam,
'O=' + out_bam,
'M=' + metrics_file,
'TMP_DIR=' + tmp_dir,
'MAX_RECORDS_IN_RAM=' + str(max_records),
])
index_bam(out_bam)
def deduplicate(in_bam, out_bam, metrics_file):
'''
Run Picard MarkDuplicates.
'''
quiet_call([
'java',
'-Xmx2g',
'-jar',
PATHS['picard'],
'MarkDuplicates',
'VALIDATION_STRINGENCY=SILENT',
'REMOVE_DUPLICATES=TRUE',
'I=' + in_bam,
'O=' + out_bam,
'M=' + metrics_file,
])
index_bam(out_bam)
def add_read_groups(in_sam, out_bam, sample_header, tmp_dir, max_records):
'''
Run Picard AddOrReplaceReadGroups.
sample_header -- set read groups to sample_header
'''
quiet_call([
PATHS['java'], '-jar',
PATHS['picard'],
'AddOrReplaceReadGroups',
'VALIDATION_STRINGENCY=SILENT',
'SO=coordinate',
'RGPL=illumina',
'RGPU=' + sample_header,
'RGSM=' + sample_header,
'RGLB=' + sample_header,
'RGID=' + sample_header,
'I=' + in_sam,
'O=' + out_bam,
'TMP_DIR=' + tmp_dir,
'MAX_RECORDS_IN_RAM=' + str(max_records),
])
def add_read_groups(in_sam, out_bam, sample_header):
'''
Run Picard AddOrReplaceReadGroups.
sample_header -- set read groups to sample_header
'''
quiet_call([
'java',
'-Xmx2g',
'-jar',
PATHS['picard'],
'AddOrReplaceReadGroups',
'VALIDATION_STRINGENCY=SILENT',
'SO=coordinate',
'RGPL=illumina',
'RGPU=' + sample_header,
'RGSM=' + sample_header,
'RGLB=' + sample_header,
'RGID=' + sample_header,
'I=' + in_sam,
'O=' + out_bam,
])
def run_haplotype_caller(bams, ref_fn, output_vcf, log_file, nct=1):
'''
Run GATK HaplotypeCaller.
'''
input_list = []
for bam in bams:
input_list.extend(('-I', bam))
quiet_call([
PATHS['java'], '-jar',
PATHS['gatk'],
'-T', 'HaplotypeCaller',
'-o', output_vcf,
'-A', 'StrandAlleleCountsBySample',
'-A', 'DepthPerSampleHC',
'-R', ref_fn,
'-nct', str(nct),
'-mmq', '5',
'-log', log_file,
'--minPruning', '0',
'--minDanglingBranchLength', '0',
'--pcr_indel_model', 'NONE',
] + input_list)
