def picard_command(config, command):
"""Returns basic AtomicJavaCmdBuilder for Picard tools commands."""
jar_path = os.path.join(config.jar_root, _PICARD_JAR)
if jar_path not in _PICARD_VERSION_CACHE:
params = AtomicJavaCmdBuilder(jar_path,
temp_root=config.temp_root,
jre_options=config.jre_options)
# Arbitrary command, since just '--version' does not work
params.set_option("MarkDuplicates")
params.set_option("--version")
requirement = versions.Requirement(
call=params.finalized_call,
name="Picard tools",
search=r"\b(\d+)\.(\d+)\.\d+",
checks=versions.GE(1, 137),
)
_PICARD_VERSION_CACHE[jar_path] = requirement
version = _PICARD_VERSION_CACHE[jar_path]
params = AtomicJavaCmdBuilder(
jar_path,
temp_root=config.temp_root,
jre_options=config.jre_options,
CHECK_JAR=version,
set_cwd=True,
)
params.set_option(command)
return params
def picard_command(config, command):
"""Returns basic AtomicJavaCmdBuilder for Picard tools commands."""
jar_path = os.path.join(config.jar_root, _PICARD_JAR)
if jar_path not in _PICARD_VERSION_CACHE:
params = AtomicJavaCmdBuilder(jar_path,
temp_root=config.temp_root,
jre_options=config.jre_options)
# Arbitrary command, since just '--version' does not work
params.set_option("MarkDuplicates")
params.set_option("--version")
requirement = versions.Requirement(call=params.finalized_call,
name="Picard tools",
search=r"\b(\d+)\.(\d+)\.\d+",
checks=versions.GE(1, 137))
_PICARD_VERSION_CACHE[jar_path] = requirement
version = _PICARD_VERSION_CACHE[jar_path]
params = AtomicJavaCmdBuilder(jar_path,
temp_root=config.temp_root,
jre_options=config.jre_options,
CHECK_JAR=version)
params.set_option(command)
return params
def test_java_builder__defaults__call():
builder = AtomicJavaCmdBuilder("/path/Foo.jar", temp_root="/disk/tmp")
assert_equal(builder.call, [
"java", "-server", "-Djava.io.tmpdir=/disk/tmp",
"-Djava.awt.headless=true", "-XX:+UseSerialGC", "-Xmx4g", "-jar",
"%(AUX_JAR)s"
])
def test_java_builder__default__no_config():
builder = AtomicJavaCmdBuilder("/path/Foo.jar")
assert_equal(builder.call, [
"java", "-server", "-Djava.io.tmpdir=%(TEMP_DIR)s",
"-Djava.awt.headless=true", "-XX:+UseSerialGC", "-Xmx4g", "-jar",
"%(AUX_JAR)s"
])
def test_java_builder__multithreaded_gc():
builder = AtomicJavaCmdBuilder("/path/Foo.jar",
temp_root="/disk/tmp",
gc_threads=3)
assert_equal(builder.call, [
"java", "-server", "-Djava.io.tmpdir=/disk/tmp",
"-Djava.awt.headless=true", "-XX:ParallelGCThreads=3", "-Xmx4g",
"-jar", "%(AUX_JAR)s"
])
def _get_gatk_version_check(config):
"""Returns a version-check object for the "GenomeAnalysisTK.jar" located at
config.jar_root; for now, this check only serves to verify that the JAR can
be executed, which may not be the case if the JRE is outdated.
"""
jar_file = os.path.join(config.jar_root, "GenomeAnalysisTK.jar")
if jar_file not in _GATK_VERSION:
params = AtomicJavaCmdBuilder(jar_file,
temp_root=config.temp_root,
jre_options=config.jre_options)
params.add_value("--version")
# Any version is fine; for now just catch old JREs
requirement = versions.Requirement(call=params.finalized_call,
name="GenomeAnalysisTK",
search=r"^(\d+)\.(\d+)",
checks=versions.Any())
_GATK_VERSION[jar_file] = requirement
return _GATK_VERSION[jar_file]
def __init__(self, config, reference, intervals, infiles, outfile,
dependencies=()):
self._basename = os.path.basename(outfile)
infiles = safe_coerce_to_tuple(infiles)
jar_file = os.path.join(config.jar_root, "GenomeAnalysisTK.jar")
command = AtomicJavaCmdBuilder(jar_file,
jre_options=config.jre_options)
command.set_option("-T", "IndelRealigner")
command.set_option("-R", "%(IN_REFERENCE)s")
command.set_option("-targetIntervals", "%(IN_INTERVALS)s")
command.set_option("-o", "%(OUT_BAMFILE)s")
command.set_option("--bam_compression", 0)
command.set_option("--disable_bam_indexing")
_set_input_files(command, infiles)
command.set_kwargs(IN_REFERENCE=reference,
IN_REF_DICT=fileutils.swap_ext(reference, ".dict"),
IN_INTERVALS=intervals,
OUT_BAMFILE=outfile,
CHECK_GATK=_get_gatk_version_check(config))
calmd = AtomicCmd(["samtools", "calmd", "-b",
"%(TEMP_IN_BAM)s", "%(IN_REF)s"],
TEMP_IN_BAM=self._basename,
IN_REF=reference,
TEMP_OUT_STDOUT=self._basename + ".calmd",
CHECK_VERSION=SAMTOOLS_VERSION)
description = "<GATK Indel Realigner (aligning): %s -> %r>" \
% (describe_files(infiles), outfile)
CommandNode.__init__(self,
description=description,
command=ParallelCmds([command.finalize(), calmd]),
dependencies=dependencies)
def __init__(self, config, reference, infiles, outfile,
threads=1, dependencies=()):
threads = _get_max_threads(reference, threads)
infiles = safe_coerce_to_tuple(infiles)
jar_file = os.path.join(config.jar_root, "GenomeAnalysisTK.jar")
command = AtomicJavaCmdBuilder(jar_file,
jre_options=config.jre_options)
command.set_option("-T", "RealignerTargetCreator")
command.set_option("-R", "%(IN_REFERENCE)s")
command.set_option("-o", "%(OUT_INTERVALS)s")
command.set_option("-nt", threads)
_set_input_files(command, infiles)
command.set_kwargs(IN_REFERENCE=reference,
IN_REF_DICT=fileutils.swap_ext(reference, ".dict"),
OUT_INTERVALS=outfile,
CHECK_GATK=_get_gatk_version_check(config))
description = "<GATK Indel Realigner (training): %s -> %r>" \
% (describe_files(infiles), outfile)
CommandNode.__init__(self,
threads=threads,
description=description,
command=command.finalize(),
dependencies=dependencies)
def test_java_builder__kwargs():
builder = AtomicJavaCmdBuilder("/path/Foo.jar", set_cwd=True)
assert_equal(builder.kwargs, {"AUX_JAR": "/path/Foo.jar", "set_cwd": True})
def test_java_builder__defaults__kwargs():
builder = AtomicJavaCmdBuilder("/path/Foo.jar")
assert_equal(builder.kwargs, {"AUX_JAR": "/path/Foo.jar"})