def customize(cls, reference, infiles, outfile, options, dependencies= ()):
assert outfile.lower().endswith('.vcf')
# Create the pileup command
pileup = AtomicCmdBuilder(
['samtools','mpileup'],
IN_REFERENCE = reference,
OUT_STDOUT = AtomicCmd.PIPE,
CHECK_SAM = SAMTOOLS_VERSION
)
pileup.set_option('-u') # uncompressed output
pileup.set_option('-r','chrUn2:1-19214051')
pileup.set_option('-f', "%(IN_REFERENCE)s") # Add reference option
for bam in infiles:
pileup.add_option(bam)
# Create variant caller command
bcftools = AtomicCmdBuilder(
['bcftools','view'],
IN_STDIN = pileup,
OUT_STDOUT = outfile
)
bcftools.set_option('-v') # output potential variant sites
bcftools.set_option('-c') # SNP calling
bcftools.set_option('-g') # call genotypes at vairant sites
bcftools.set_option('-') # STDIN
return {
"commands" : {
"pileup" : pileup,
"bcftools" : bcftools,
}
}
def customize(cls, reference, infile, outfile, filters, options, dependencies = ()):
# filter reads
percentile = str(options.makefile['vcf_percentile_threshold'])
flt = AtomicCmdBuilder(['vcf_qual_percentile'],
IN_VCF = infile,
OUT_VCF = outfile
)
for key,val in filters.items():
flt.add_option(key,val)
flt.set_option('--out','%(OUT_VCF)s')
flt.add_option(infile)
return {
'commands':{
'Filter': flt
}
}
def test_builder__finalize__calls_atomiccmd():
was_called = []
class _AtomicCmdMock:
def __init__(self, *args, **kwargs):
assert_equal(args, (["echo", "-out", "%(OUT_FILE)s", "%(IN_FILE)s"],))
assert_equal(kwargs, {"IN_FILE" : "/in/file", "OUT_FILE" : "/out/file", "set_cwd" : True})
was_called.append(True)
with Monkeypatch("pypeline.atomiccmd.builder.AtomicCmd", _AtomicCmdMock):
builder = AtomicCmdBuilder("echo", set_cwd = True)
builder.add_option("-out", "%(OUT_FILE)s")
builder.add_value("%(IN_FILE)s")
builder.set_kwargs(OUT_FILE = "/out/file",
IN_FILE = "/in/file")
builder.finalize()
assert was_called
def test_builder__add_option__overwrite():
builder = AtomicCmdBuilder("find")
builder.add_option("-name", "*.txt")
builder.add_option("-or")
builder.add_option("-name", "*.bat")
assert_equal(builder.call,
["find", "-name", "*.txt", "-or", "-name", "*.bat"])
def test_builder__finalize__calls_atomiccmd():
was_called = []
class _AtomicCmdMock:
def __init__(self, *args, **kwargs):
assert_equal(args,
(["echo", "-out", "%(OUT_FILE)s", "%(IN_FILE)s"], ))
assert_equal(kwargs, {
"IN_FILE": "/in/file",
"OUT_FILE": "/out/file",
"set_cwd": True
})
was_called.append(True)
with Monkeypatch("pypeline.atomiccmd.builder.AtomicCmd", _AtomicCmdMock):
builder = AtomicCmdBuilder("echo", set_cwd=True)
builder.add_option("-out", "%(OUT_FILE)s")
builder.add_value("%(IN_FILE)s")
builder.set_kwargs(OUT_FILE="/out/file", IN_FILE="/in/file")
builder.finalize()
assert was_called
def customize(cls, groups, prefix, options, dependencies = ()):
# Merge the VCF files
merge_vcf = AtomicCmdBuilder(['vcf_merge'], OUT_VCF = "merged.vcf")
for group in groups:
vcf_file = os.path.join(options.makefile['RecalDir'],
'gatk.{}.{}.raw.recal_final.vcf'.format(group,prefix)
)
merge_vcf.add_option("-i",vcf_file)
merge_vcf.add_option("-o", '%(OUT_VCF)s')
# Create the snp list
snp_list = AtomicCmdBuilder(['vcf_snp_list'])
snp_list.add_option('--recal_dir',options.makefile['RecalDir'])
snp_list.add_option('')
return {
'commands' : {
'merge' : merge_vcf,
'Snp' : snp_list
}
}
def test_builder__pop_option__last_option():
builder = AtomicCmdBuilder("find")
builder.add_option("-size", "0", fixed=False)
builder.add_option("-size", "1", fixed=False)
builder.pop_option("-size")
assert_equal(builder.call, ["find", "-size", "0"])
def test_builder__pop_option__last_option():
builder = AtomicCmdBuilder("find")
builder.add_option("-size", "0", fixed = False)
builder.add_option("-size", "1", fixed = False)
builder.pop_option("-size")
assert_equal(builder.call, ["find", "-size", "0"])
def test_builder__add_option__overwrite():
builder = AtomicCmdBuilder("find")
builder.add_option("-name", "*.txt")
builder.add_option("-or")
builder.add_option("-name", "*.bat")
assert_equal(builder.call, ["find", "-name", "*.txt", "-or", "-name", "*.bat"])
