def run_vardict(): params.v = True params.G = ref params.b = infile params._stdout = outfile Shell().vardict(**params).run() if gz: shell.gzip(outfile)
def run_elprep(): elpsh = Shell(subcmd = True, equal = ' ').elprep params['log-path'] = joboutdir params['nr-of-threads'] = nthread params['sorting-order'] = sortby if steps.sort else 'keep' params['mark-duplicates'] = steps.markdup params['remove-duplicates'] = steps.rmdup params[''] = [infile, outfile] if steps.markdup: params['mark-optical-duplicates'] = path.join(joboutdir, inprefix + '.opticaldups.txt') if steps.recal: params['bqsr'] = path.join(joboutdir, inprefix + '.bqsr.txt') params['bqsr-reference'] = ref + '.elprep' if knownSites: params['known-sites'] = knownSites elpsh.filter(**params).run() if steps.index: subshell.samtools.index(outfile, outfile + '.bai')
def run_picard():
global infile
mem = mem2(argsmem, '-jdict')
mem['-Djava.io.tmpdir'] = tmpdir
shellpicard = Shell(subcmd = True, dash = '', equal = '=').picard(**mem)
if not (steps.sort or steps.index or steps.markdup or steps.rmdup):
shellpicard.SamFormatConverter(TMP_DIR = tmpdir, I = infile, O = outfile).run()
else:
bamfile = outfile
if steps.sort:
bamfile = path.join(joboutdir, inprefix + '.sorted.bam')
shellpicard.ShortSam(TMP_DIR = tmpdir, I = infile, O = bamfile, SO = sortby).run()
if infile != {{i.infile | quote}}:
shell.rm(f = True, _ = infile)
infile = bamfile
if steps.markdup:
mfile = "/dev/null"
bamfile = path.join(joboutdir, inprefix + '.dedup.bam')
shellpicard.MarkDuplicates(REMOVE_DUPLICATES = 'true' if steps.rmdup else 'false', TMP_DIR = tmpdir, I = infile, O = bamfile, M = mfile).run()
if infile != {{i.infile | quote}}:
shell.rm(f = True, _ = infile)
infile = bamfile
if steps.index:
shellpicard.BuildBamIndex(TMP_DIR = tmpdir, I = infile, O = outfile + '.bai').run()
if infile != outfile:
if path.exists(infile + '.bai'):
shell.mv(infile + '.bai', outfile + '.bai')
shell.mv(infile, outfile)
def run_gatk():
gatkmem = mem2(mem, 'jdict')
gatkmem['Djava.io.tmpdir={!r}'.format(tmpdir)] = True
gatksh = Shell(equal = ' ', dash = '-')
params.T = 'HaplotypeCaller'
params.R = ref
params.I = infile
params.o = outfile
params.nct = nthread
params._ = list(gatkmem.keys())
gatksh(**params).run()
if gz: shell.gzip(outfile)
def run_biobambam(): mem = mem2(argsmem, 'M') if steps.index: params.index = 1 params.indexfilename = outfile + '.bai' params.I = infile params.O = outfile params.SO = sortby params.blockme = mem params.tmpfile = path.join(tmpdir, 'biobambam.tmp') params.inputformat = infmt params.outfmt = 'bam' params.inputthreads = nthread params.outputthreads = nthread params.markduplicates = int(steps.markdup) params.rmdup = int(steps.rmdup) Shell(dash = '', equal = '=').biobambam(**params).run()
def run_samtools():
shellst = Shell(subcmd=True).samtools
params['@'] = nthread
shellst.index(_=[infile, outidx]).run()
shell.TOOLS.update(dict(
biobambam = biobambam,
sambamba = sambamba,
samtools = samtools,
picard = picard,
elprep = elprep
))
if steps.rmdup:
steps.markdup = True
if not path.exists(tmpdir):
shell.mkdir(tmpdir, p = True)
if steps.recal and tool != 'elprep':
raise ValueError('Step "recal" is only enabled by "elprep", use pBamRecal for other tools.')
subshell = Shell(subcmd = True)
def run_biobambam():
mem = mem2(argsmem, 'M')
if steps.index:
params.index = 1
params.indexfilename = outfile + '.bai'
params.I = infile
params.O = outfile
params.SO = sortby
params.blockme = mem
params.tmpfile = path.join(tmpdir, 'biobambam.tmp')
params.inputformat = infmt
params.outfmt = 'bam'
params.inputthreads = nthread
