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 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, input_file, k_groups, output_root,
samples=None, dependencies=()):
self._samples = samples
self._input_file = input_file
self._k_groups = k_groups
group_key = "Group(%i)" % (self._k_groups,)
self._supervised = samples and any((row[group_key] != '-')
for row in samples.itervalues())
assert k_groups in (2, 3), k_groups
prefix = os.path.splitext(os.path.basename(input_file))[0]
output_prefix = os.path.join(output_root,
"%s.%i" % (prefix, k_groups))
cmd = AtomicCmdBuilder("admixture",
IN_FILE_BED=input_file,
IN_FILE_BIM=fileutils.swap_ext(input_file,
".bim"),
IN_FILE_FAM=fileutils.swap_ext(input_file,
".fam"),
TEMP_OUT_FILE_BED=prefix + ".bed",
TEMP_OUT_FILE_BIM=prefix + ".bim",
TEMP_OUT_FILE_FAM=prefix + ".fam",
TEMP_OUT_FILE_POP=prefix + ".pop",
OUT_P=output_prefix + ".P",
OUT_Q=output_prefix + ".Q",
OUT_STDOUT=output_prefix + ".log",
CHECK_VERSION=ADMIXTURE_VERSION,
set_cwd=True)
cmd.set_option("-s", random.randint(0, 2 ** 16 - 1))
if self._supervised:
cmd.set_option("--supervised")
cmd.add_value("%(TEMP_OUT_FILE_BED)s")
cmd.add_value(int(k_groups))
CommandNode.__init__(self,
description="<Admixture -> '%s.*''>"
% (output_prefix,),
command=cmd.finalize(),
dependencies=dependencies)
def __init__(self, input_file, k_groups, output_root,
samples=None, dependencies=()):
self._samples = samples
self._input_file = input_file
self._k_groups = k_groups
group_key = "Group(%i)" % (self._k_groups,)
self._supervised = samples and any((row[group_key] != '-')
for row in samples.itervalues())
assert k_groups in (2, 3), k_groups
prefix = os.path.splitext(os.path.basename(input_file))[0]
output_prefix = os.path.join(output_root,
"%s.%i" % (prefix, k_groups))
cmd = AtomicCmdBuilder("admixture",
IN_FILE_BED=input_file,
IN_FILE_BIM=fileutils.swap_ext(input_file,
".bim"),
IN_FILE_FAM=fileutils.swap_ext(input_file,
".fam"),
TEMP_OUT_FILE_BED=prefix + ".bed",
TEMP_OUT_FILE_BIM=prefix + ".bim",
TEMP_OUT_FILE_FAM=prefix + ".fam",
TEMP_OUT_FILE_POP=prefix + ".pop",
OUT_P=output_prefix + ".P",
OUT_Q=output_prefix + ".Q",
OUT_STDOUT=output_prefix + ".log",
CHECK_VERSION=ADMIXTURE_VERSION,
set_cwd=True)
cmd.set_option("-s", random.randint(0, 2 ** 16 - 1))
if self._supervised:
cmd.set_option("--supervised")
cmd.add_value("%(TEMP_OUT_FILE_BED)s")
cmd.add_value(int(k_groups))
CommandNode.__init__(self,
description="<Admixture -> '%s.*''>"
% (output_prefix,),
command=cmd.finalize(),
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 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("paleomix.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,
input_file,
k_groups,
output_root,
groups,
dependencies=()):
self._groups = groups
self._input_file = input_file
prefix = os.path.splitext(os.path.basename(input_file))[0]
output_prefix = os.path.join(output_root, "%s.%i" % (prefix, k_groups))
cmd = AtomicCmdBuilder(
"admixture",
IN_FILE_BED=input_file,
IN_FILE_BIM=fileutils.swap_ext(input_file, ".bim"),
IN_FILE_FAM=fileutils.swap_ext(input_file, ".fam"),
TEMP_OUT_FILE_BED=prefix + ".bed",
TEMP_OUT_FILE_BIM=prefix + ".bim",
TEMP_OUT_FILE_FAM=prefix + ".fam",
TEMP_OUT_FILE_POP=prefix + ".pop",
OUT_P=output_prefix + ".P",
OUT_Q=output_prefix + ".Q",
OUT_STDOUT=output_prefix + ".log",
CHECK_VERSION=ADMIXTURE_VERSION,
set_cwd=True,
)
cmd.set_option("-s", random.randint(0, 2**16 - 1))
cmd.set_option("--supervised")
cmd.add_value("%(TEMP_OUT_FILE_BED)s")
cmd.add_value(int(k_groups))
CommandNode.__init__(
self,
description="<Admixture -> '%s.*''>" % (output_prefix, ),
command=cmd.finalize(),
dependencies=dependencies,
)
def test_builder__finalize__calls_atomiccmd():
was_called = []
class _AtomicCmdMock(object):
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("paleomix.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_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"])