def customize(cls, reference, infile, outfile, regions = None, dependencies = ()):
assert outfile.lower().endswith(".vcf.bgz")
pileup = AtomicCmdBuilder(["samtools", "mpileup"],
IN_REFERENCE = reference,
IN_BAMFILE = infile,
IN_REGIONS = regions,
OUT_STDOUT = AtomicCmd.PIPE,
CHECK_SAM = SAMTOOLS_VERSION)
pileup.set_option("-u") # Uncompressed output
pileup.set_option("-f", "%(IN_REFERENCE)s")
pileup.add_value("%(IN_BAMFILE)s")
if regions:
pileup.set_option("-l", "%(IN_REGIONS)s")
genotype = AtomicCmdBuilder(["bcftools", "view"],
IN_STDIN = pileup,
OUT_STDOUT = AtomicCmd.PIPE)
genotype.add_value("-")
bgzip = AtomicCmdBuilder(["bgzip"],
IN_STDIN = genotype,
OUT_STDOUT = outfile)
return {"commands" : {"pileup" : pileup,
"genotype" : genotype,
"bgzip" : bgzip}}
def customize(cls, input_file, output_file, algorithm = "auto", dependencies = ()):
command = AtomicCmdBuilder(_PRESETS[algorithm.lower()])
command.add_value("%(IN_FASTA)s")
command.set_kwargs(IN_FASTA = input_file,
OUT_STDOUT = output_file,
CHECK_VERSION = MAFFT_VERSION)
return {"command" : command,
"dependencies" : dependencies}
def __init__(self,
config,
reference,
input_bam,
output_bam,
tags,
min_mapq=0,
filter_unmapped=False,
dependencies=()):
flt_params = AtomicCmdBuilder(("samtools", "view", "-bu"),
IN_BAM=input_bam,
OUT_STDOUT=AtomicCmd.PIPE)
if min_mapq:
flt_params.set_option("-q", min_mapq, sep="")
if filter_unmapped:
flt_params.set_option("-F", "0x4", sep="")
flt_params.add_value("%(IN_BAM)s")
jar_params = picard.picard_command(config, "AddOrReplaceReadGroups")
jar_params.set_option("INPUT", "/dev/stdin", sep="=")
# Output is written to a named pipe, since the JVM may, in some cases,
# emit warning messages to stdout, resulting in a malformed BAM.
jar_params.set_option("OUTPUT", "%(TEMP_OUT_BAM)s", sep="=")
jar_params.set_option("COMPRESSION_LEVEL", "0", sep="=")
# Ensure that the BAM is sorted; this is required by the pipeline, and
# needs to be done before calling calmd (avoiding pathologic runtimes).
jar_params.set_option("SORT_ORDER", "coordinate", sep="=")
# All tags are overwritten; ID is set since the default (e.g. '1')
# causes problems with pysam due to type inference (is read as a length
# 1 string, but written as a character).
for tag in ("ID", "SM", "LB", "PU", "PL"):
jar_params.set_option(tag, tags[tag], sep="=")
jar_params.set_kwargs(IN_STDIN=flt_params, TEMP_OUT_BAM="bam.pipe")
calmd = AtomicCmdBuilder(
["samtools", "calmd", "-b", "%(TEMP_IN_BAM)s", "%(IN_REF)s"],
IN_REF=reference,
TEMP_IN_BAM="bam.pipe",
OUT_STDOUT=output_bam)
commands = [cmd.finalize() for cmd in (flt_params, jar_params, calmd)]
description = "<Cleanup BAM: %s -> '%s'>" \
% (input_bam, output_bam)
PicardNode.__init__(self,
command=ParallelCmds(commands),
description=description,
dependencies=dependencies)
def customize(cls, reference, in_bam, in_vcf, outfile, dependencies=()):
unicat = AtomicCmdBuilder(["unicat", "%(IN_VCF)s"],
IN_VCF=in_vcf,
OUT_STDOUT=AtomicCmd.PIPE)
vcfpileup = AtomicCmdBuilder(["vcf_create_pileup", "%(OUT_PILEUP)s"],
IN_REF=reference,
IN_BAM=in_bam,
IN_STDIN=unicat,
OUT_PILEUP=outfile,
OUT_TBI=outfile + ".tbi")
vcfpileup.add_value("%(IN_BAM)s")
vcfpileup.set_option("-f", "%(IN_REF)s")
return {"commands": {"unicat": unicat, "pileup": vcfpileup}}
def customize(cls, reference, in_bam, in_vcf, outfile, dependencies = ()):
unicat = AtomicCmdBuilder(["unicat", "%(IN_VCF)s"],
IN_VCF = in_vcf,
OUT_STDOUT = AtomicCmd.PIPE)
vcfpileup = AtomicCmdBuilder(["vcf_create_pileup", "%(OUT_PILEUP)s"],
IN_REF = reference,
IN_BAM = in_bam,
IN_STDIN = unicat,
OUT_PILEUP = outfile,
OUT_TBI = outfile + ".tbi")
vcfpileup.add_value("%(IN_BAM)s")
vcfpileup.set_option("-f", "%(IN_REF)s")
return {"commands" : {"unicat" : unicat,
"pileup" : vcfpileup}}
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 __init__(self, config, reference, input_bam, output_bam, tags,
min_mapq=0, filter_unmapped=False, dependencies=()):
flt_params = AtomicCmdBuilder(("samtools", "view", "-bu"),
IN_BAM=input_bam,
OUT_STDOUT=AtomicCmd.PIPE)
if min_mapq:
flt_params.set_option("-q", min_mapq, sep="")
if filter_unmapped:
flt_params.set_option("-F", "0x4", sep="")
flt_params.add_value("%(IN_BAM)s")
jar_params = picard.picard_command(config, "AddOrReplaceReadGroups")
jar_params.set_option("INPUT", "/dev/stdin", sep="=")
# Output is written to a named pipe, since the JVM may, in some cases,
# emit warning messages to stdout, resulting in a malformed BAM.
jar_params.set_option("OUTPUT", "%(TEMP_OUT_BAM)s", sep="=")
jar_params.set_option("COMPRESSION_LEVEL", "0", sep="=")
# Ensure that the BAM is sorted; this is required by the pipeline, and
# needs to be done before calling calmd (avoiding pathologic runtimes).
jar_params.set_option("SORT_ORDER", "coordinate", sep="=")
# All tags are overwritten; ID is set since the default (e.g. '1')
# causes problems with pysam due to type inference (is read as a length
# 1 string, but written as a character).
for tag in ("ID", "SM", "LB", "PU", "PL"):
jar_params.set_option(tag, tags[tag], sep="=")
jar_params.set_kwargs(IN_STDIN=flt_params,
TEMP_OUT_BAM="bam.pipe")
calmd = AtomicCmdBuilder(["samtools", "calmd", "-b",
"%(TEMP_IN_BAM)s", "%(IN_REF)s"],
IN_REF=reference,
TEMP_IN_BAM="bam.pipe",
OUT_STDOUT=output_bam)
commands = [cmd.finalize() for cmd in (flt_params, jar_params, calmd)]
description = "<Cleanup BAM: %s -> '%s'>" \
% (input_bam, output_bam)
PicardNode.__init__(self,
command=ParallelCmds(commands),
description=description,
dependencies=dependencies)
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 __init__(self, d_bam, dependencies):
aux_r = os.path.join(os.path.dirname(gccorrect.__file__), 'model_gc.R')
dest = os.path.join(d_bam.bam_output,
'%s_GC_Model.txt' % (str(d_bam.bam_name)))
plot_dest = os.path.splitext(dest)[0] + '.pdf'
infiles = [''.join(n.output_files) for n in dependencies]
builder = AtomicCmdBuilder(("Rscript", "%(AUX_R)s"))
builder.add_value('%(OUT_FILEPATH)s')
builder.add_value('%(OUT_PLOT)s')
builder.add_multiple_values(infiles)
builder.set_kwargs(AUX_R=aux_r,
OUT_FILEPATH=dest,
OUT_PLOT=plot_dest,
CHECK_VERSION=Rscript_VERSION)
cmd = builder.finalize()
d_bam.opts['BamInfo']['--GCmodel'] = dest
description = "<CreateGCModel: '%s' -> '%s'" % (d_bam.bam_temp_local,
dest)
CommandNode.__init__(self,
description=description,
command=cmd,
dependencies=dependencies)
def _process_output(stdin, output_file, reference, run_fixmate = False):
convert = AtomicCmdBuilder("safeSAM2BAM")
convert.set_option("--flag-as-sorted")
convert.set_option("-F", "0x4", sep = "", fixed = False) # Remove misses
convert.set_kwargs(IN_STDIN = stdin,
OUT_STDOUT = AtomicCmd.PIPE,
CHECK_PYSAM = PYSAM_VERSION,
CHECK_SAMTOOLS = SAMTOOLS_VERSION)
fixmate = None
if run_fixmate:
fixmate = AtomicCmdBuilder(("samtools", "fixmate", "-", "-"),
IN_STDIN = convert,
OUT_STDOUT = AtomicCmd.PIPE,
CHECK_SAMTOOLS = SAMTOOLS_VERSION)
sort = AtomicCmdBuilder(("samtools", "sort"))
sort.set_option("-o") # Output to STDOUT on completion
sort.add_value("-")
sort.add_value("%(TEMP_OUT_BAM)s")
sort.set_kwargs(IN_STDIN = fixmate or convert,
OUT_STDOUT = AtomicCmd.PIPE,
TEMP_OUT_BAM = "sorted",
CHECK_SAM = SAMTOOLS_VERSION)
calmd = AtomicCmdBuilder(("samtools", "calmd"))
calmd.add_value("-")
calmd.add_value("%(IN_REF)s")
calmd.set_option("-b") # Output BAM
calmd.set_kwargs(IN_REF = reference,
IN_STDIN = sort,
OUT_STDOUT = output_file,
CHECK_SAM = SAMTOOLS_VERSION)
order = ["convert", "sort", "calmd"]
commands = {"convert" : convert,
"sort" : sort,
"calmd" : calmd}
if run_fixmate:
order.insert(1, "fixmate")
commands["fixmate"] = fixmate
return order, commands
def _process_output(stdin, output_file, reference, run_fixmate=False):
convert = AtomicCmdBuilder("safeSAM2BAM")
convert.set_option("--flag-as-sorted")
convert.set_option("-F", "0x4", sep="", fixed=False) # Remove misses
convert.set_kwargs(IN_STDIN=stdin,
OUT_STDOUT=AtomicCmd.PIPE,
CHECK_PYSAM=PYSAM_VERSION,
CHECK_SAMTOOLS=SAMTOOLS_VERSION)
fixmate = None
if run_fixmate:
fixmate = AtomicCmdBuilder(("samtools", "fixmate", "-", "-"),
IN_STDIN=convert,
OUT_STDOUT=AtomicCmd.PIPE,
CHECK_SAMTOOLS=SAMTOOLS_VERSION)
sort = AtomicCmdBuilder(("samtools", "sort"))
sort.set_option("-o") # Output to STDOUT on completion
sort.add_value("-")
sort.add_value("%(TEMP_OUT_BAM)s")
sort.set_kwargs(IN_STDIN=fixmate or convert,
OUT_STDOUT=AtomicCmd.PIPE,
TEMP_OUT_BAM="sorted",
CHECK_SAM=SAMTOOLS_VERSION)
calmd = AtomicCmdBuilder(("samtools", "calmd"))
calmd.add_value("-")
calmd.add_value("%(IN_REF)s")
calmd.set_option("-b") # Output BAM
calmd.set_kwargs(IN_REF=reference,
IN_STDIN=sort,
OUT_STDOUT=output_file,
CHECK_SAM=SAMTOOLS_VERSION)
order = ["convert", "sort", "calmd"]
commands = {"convert": convert, "sort": sort, "calmd": calmd}
if run_fixmate:
order.insert(1, "fixmate")
commands["fixmate"] = fixmate
return order, commands
def test_builder__add_value__two_values():
builder = AtomicCmdBuilder("ls")
builder.add_value("%(IN_FILE)s")
builder.add_value("%(OUT_FILE)s")
assert_equal(builder.call, ["ls", "%(IN_FILE)s", "%(OUT_FILE)s"])
def test_builder__add_value__two_values():
builder = AtomicCmdBuilder("ls")
builder.add_value("%(IN_FILE)s")
builder.add_value("%(OUT_FILE)s")
assert_equal(builder.call, ["ls", "%(IN_FILE)s", "%(OUT_FILE)s"])
