def preprocess(self, preargs):
self.base = preargs[0]
self.fwd = preargs[1]
self.rev = preargs[2]
self.datadir = preargs[3]
#Run quality control ananlysis and quality based trimming
qcdir = os.path.abspath(self.datadir) + '/qc'
if not os.path.exists(qcdir):
os.mkdir(qcdir)
self.fwd, self.rev = trimgalore.run_qc(self.fwd, self.rev, qcdir,
self.base)
#Run alignment
self.bamfile = bowtie.bowtie(self.fwd, self.rev, self.datadir,
self.base)
#Run picard tools
self.bamfile = picard.addreadgroup(self.bamfile, self.base)
self.bamfile, self.metrics = picard.markdup(self.bamfile)
#Run GATK realignment and quality recalibration
self.interval = gatk.target(self.bamfile)
self.bamfile = gatk.realigner(self.bamfile, self.interval)
self.table = gatk.baserecal(self.bamfile)
self.bamfile = gatk.printreads(self.bamfile, self.table)
self.vcf = gatk.haplocaller(self.bamfile)
self.srf, self.irf, self.stf, self.itf = gatk.vqsr(self.vcf, self.outdir)
self.vcffile = gatk.applyrecalibration(self.srf, self.irf, self.stf,
self.itf, self.vcffile, self.outdir)
return(self.vcffile)
def vqsr(self, vcffile):
#Run VQSR and apply reclibration
self.vcffile = vcffile
self.srf, self.irf, self.stf, self.itf = gatk.vqsr(vcffile, self.outdir)
self.vcffile = gatk.applyrecalibration(self.srf, self.irf, self.stf,
self.itf, self.vcffile, self.outdir)
return(self.vcffile)
def preprocess(self, preargs):
self.base = preargs[0]
self.fwd = preargs[1]
self.rev = preargs[2]
self.datadir = preargs[3]
#Run quality control ananlysis and quality based trimming
qcdir = os.path.abspath(self.datadir) + '/qc'
if not os.path.exists(qcdir):
os.mkdir(qcdir)
self.fwd, self.rev = trimgalore.run_qc(self.fwd, self.rev, qcdir,
self.base)
#Run alignment
self.bamfile = bowtie.bowtie(self.fwd, self.rev, self.datadir,
self.base)
#Run picard tools
self.bamfile = picard.addreadgroup(self.bamfile, self.base)
self.bamfile, self.metrics = picard.markdup(self.bamfile)
#Run GATK realignment and quality recalibration
self.interval = gatk.target(self.bamfile)
self.bamfile = gatk.realigner(self.bamfile, self.interval)
self.table = gatk.baserecal(self.bamfile)
self.bamfile = gatk.printreads(self.bamfile, self.table)
self.vcf = gatk.haplocaller(self.bamfile)
self.srf, self.irf, self.stf, self.itf = gatk.vqsr(
self.vcf, self.outdir)
self.vcffile = gatk.applyrecalibration(self.srf, self.irf, self.stf,
self.itf, self.vcffile,
self.outdir)
return (self.vcffile)