def snp_and_indel_bcf(self):
jobs = []
input_bams = [os.path.join("alignment", sample.name, sample.name + ".sorted.dup.recal.bam") for sample in self.samples]
nb_jobs = config.param('snp_and_indel_bcf', 'approxNbJobs', required=False, type='int')
output_directory = "variants/rawBCF"
bcftools_view_options = "-bvcg"
if nb_jobs and nb_jobs > 1:
for region in self.generate_approximate_windows(nb_jobs):
job = pipe_jobs([
samtools.mpileup(input_bams, None, config.param('snp_and_indel_bcf', 'extra_mpileup_options'), region),
samtools.bcftools_view("-", os.path.join(output_directory, "allSamples." + region + ".bcf"), bcftools_view_options),
])
job.name = "snp_and_indel_bcf.allSamples." + re.sub(":", "_", region)
jobs.append(job)
else:
job = pipe_jobs([
samtools.mpileup(input_bams, None, config.param('snp_and_indel_bcf', 'extra_mpileup_options')),
samtools.bcftools_view("-", os.path.join(output_directory, "allSamples.bcf"), bcftools_view_options),
])
job.name = "snp_and_indel_bcf.allSamples"
jobs.append(job)
for job in jobs:
job.command = "mkdir -p " + output_directory + " && \\\n" + job.command
return jobs
def merge_filter_bcf(self):
nb_jobs = config.param('snp_and_indel_bcf', 'approxNbJobs', type='int')
inputs = ["variants/rawBCF/allSamples." + region + ".bcf" for region in self.generate_approximate_windows(nb_jobs)]
output_file_prefix = "variants/allSamples.merged."
bcf = output_file_prefix + "bcf"
job = concat_jobs([
samtools.bcftools_cat(inputs, bcf),
samtools.bcftools_view(bcf, output_file_prefix + "flt.vcf")
])
job.name = "merge_filter_bcf"
return [job]