def _gatk4_cmd(jvm_opts, params, data):
"""Retrieve unified command for GATK4, using 'gatk'. GATK3 is 'gatk3'.
"""
gatk_cmd = utils.which(os.path.join(os.path.dirname(os.path.realpath(sys.executable)), "gatk"))
return "%s && export PATH=%s:\"$PATH\" && gatk --java-options '%s' %s" % \
(utils.clear_java_home(), utils.get_java_binpath(gatk_cmd),
" ".join(jvm_opts), " ".join([str(x) for x in params]))
def run(self, subcmd, opts, memscale=None):
jvm_opts = get_picard_opts(self._config, memscale=memscale)
cmd = ["export", "PATH=%s:$PATH" % utils.get_java_binpath(), "&&"] + \
[self._cmd] + jvm_opts + [subcmd] + ["%s=%s" % (x, y) for x, y in opts] + \
["VALIDATION_STRINGENCY=SILENT"]
do.run(utils.clear_java_home() + " && " + " ".join(cmd),
"Picard: %s" % subcmd)
def _gatk4_cmd(jvm_opts, params, data):
"""Retrieve unified command for GATK4, using 'gatk'. GATK3 is 'gatk3'.
"""
gatk_cmd = utils.which(os.path.join(os.path.dirname(os.path.realpath(sys.executable)), "gatk"))
return "%s && export PATH=%s:\"$PATH\" && gatk --java-options '%s' %s" % \
(utils.clear_java_home(), utils.get_java_binpath(gatk_cmd),
" ".join(jvm_opts), " ".join([str(x) for x in params]))
def _get_snpeff_cmd(cmd_name, datadir, data, out_file):
"""Retrieve snpEff base command line.
"""
resources = config_utils.get_resources("snpeff", data["config"])
jvm_opts = resources.get("jvm_opts", ["-Xms750m", "-Xmx3g"])
# scale by cores, defaulting to 2x base usage to ensure we have enough memory
# for single core runs to use with human genomes.
# Sets a maximum amount of memory to avoid core dumps exceeding 32Gb
# We shouldn't need that much memory for snpEff, so avoid issues
# https://www.elastic.co/guide/en/elasticsearch/guide/current/heap-sizing.html#compressed_oops
jvm_opts = config_utils.adjust_opts(
jvm_opts, {
"algorithm": {
"memory_adjust": {
"direction": "increase",
"maximum": "30000M",
"magnitude": max(2, dd.get_cores(data))
}
}
})
memory = " ".join(jvm_opts)
snpeff = config_utils.get_program("snpEff", data["config"])
java_args = "-Djava.io.tmpdir=%s" % utils.safe_makedir(
os.path.join(os.path.dirname(out_file), "tmp"))
export = "unset JAVA_HOME && export PATH=%s:$PATH && " % (
utils.get_java_binpath())
cmd = "{export} {snpeff} {memory} {java_args} {cmd_name} -dataDir {datadir}"
return cmd.format(**locals())
def _gatk4_cmd(jvm_opts, params, data):
"""Retrieve unified command for GATK4, using gatk-launch.
"""
gatk_cmd = utils.which(
os.path.join(os.path.dirname(os.path.realpath(sys.executable)),
"gatk-launch"))
return "unset JAVA_HOME && export PATH=%s:$PATH && gatk-launch --javaOptions '%s' %s" % \
(utils.get_java_binpath(gatk_cmd), " ".join(jvm_opts), " ".join([str(x) for x in params]))
def _get_snpeff_cmd(cmd_name, datadir, data, out_file):
"""Retrieve snpEff base command line.
"""
resources = config_utils.get_resources("snpeff", data["config"])
jvm_opts = resources.get("jvm_opts", ["-Xms750m", "-Xmx3g"])
# scale by cores, defaulting to 2x base usage to ensure we have enough memory
# for single core runs to use with human genomes
jvm_opts = config_utils.adjust_opts(jvm_opts, {"algorithm": {"memory_adjust":
{"direction": "increase",
"magnitude": max(2, dd.get_cores(data))}}})
memory = " ".join(jvm_opts)
snpeff = config_utils.get_program("snpEff", data["config"])
java_args = "-Djava.io.tmpdir=%s" % utils.safe_makedir(os.path.join(os.path.dirname(out_file), "tmp"))
export = "unset JAVA_HOME && export PATH=%s:$PATH && " % (utils.get_java_binpath())
cmd = "{export} {snpeff} {memory} {java_args} {cmd_name} -dataDir {datadir}"
return cmd.format(**locals())
def _get_snpeff_cmd(cmd_name, datadir, data, out_file):
"""Retrieve snpEff base command line.
"""
resources = config_utils.get_resources("snpeff", data["config"])
jvm_opts = resources.get("jvm_opts", ["-Xms750m", "-Xmx3g"])
# scale by cores, defaulting to 2x base usage to ensure we have enough memory
# for single core runs to use with human genomes
jvm_opts = config_utils.adjust_opts(jvm_opts, {"algorithm": {"memory_adjust":
{"direction": "increase",
"magnitude": max(2, dd.get_cores(data))}}})
memory = " ".join(jvm_opts)
snpeff = config_utils.get_program("snpEff", data["config"])
java_args = "-Djava.io.tmpdir=%s" % utils.safe_makedir(os.path.join(os.path.dirname(out_file), "tmp"))
export = "unset JAVA_HOME && export PATH=%s:$PATH && " % (utils.get_java_binpath())
cmd = "{export} {snpeff} {memory} {java_args} {cmd_name} -dataDir {datadir}"
return cmd.format(**locals())
def gatk_cmd(name, jvm_opts, params, config=None):
"""Retrieve PATH to gatk using locally installed java.
"""
if name == "gatk":
if isinstance(config, dict) and "config" not in config:
data = {"config": config}
else:
data = config
if not data or "gatk4" not in dd.get_tools_off(data):
return _gatk4_cmd(jvm_opts, params, data)
gatk_cmd = utils.which(
os.path.join(os.path.dirname(os.path.realpath(sys.executable)), name))
# if we can't find via the local executable, fallback to being in the path
if not gatk_cmd:
gatk_cmd = utils.which(name)
if gatk_cmd:
return "%s && export PATH=%s:$PATH && %s %s %s" % \
(utils.clear_java_home(), utils.get_java_binpath(gatk_cmd), gatk_cmd,
" ".join(jvm_opts), " ".join([str(x) for x in params]))
def gatk_cmd(name, jvm_opts, params, config=None):
"""Retrieve PATH to gatk or gatk-framework executable using locally installed java.
"""
if name == "gatk":
assert config, "Need configuration input for gatk to distinguish gatk4"
if isinstance(config, dict) and "config" not in config:
data = {"config": config}
else:
data = config
if "gatk4" in dd.get_tools_on(data):
return _gatk4_cmd(jvm_opts, params, data)
gatk_cmd = utils.which(os.path.join(os.path.dirname(os.path.realpath(sys.executable)), name))
# if we can't find via the local executable, fallback to being in the path
if not gatk_cmd:
gatk_cmd = utils.which(name)
if gatk_cmd:
return "unset JAVA_HOME && export PATH=%s:$PATH && %s %s %s" % \
(utils.get_java_binpath(gatk_cmd), gatk_cmd,
" ".join(jvm_opts), " ".join([str(x) for x in params]))
def gatk_cmd(name, jvm_opts, params, config=None):
"""Retrieve PATH to gatk using locally installed java.
"""
if name == "gatk":
if isinstance(config, dict) and "config" not in config:
data = {"config": config}
else:
data = config
if not data or "gatk4" not in dd.get_tools_off(data):
return _gatk4_cmd(jvm_opts, params, data)
else:
name = "gatk3"
gatk_cmd = utils.which(os.path.join(os.path.dirname(os.path.realpath(sys.executable)), name))
# if we can't find via the local executable, fallback to being in the path
if not gatk_cmd:
gatk_cmd = utils.which(name)
if gatk_cmd:
return "%s && export PATH=%s:\"$PATH\" && %s %s %s" % \
(utils.clear_java_home(), utils.get_java_binpath(gatk_cmd), gatk_cmd,
" ".join(jvm_opts), " ".join([str(x) for x in params]))
def gatk_cmd(name, jvm_opts, params, config=None):
"""Retrieve PATH to gatk or gatk-framework executable using locally installed java.
"""
if name == "gatk":
assert config, "Need configuration input for gatk to distinguish gatk4"
if isinstance(config, dict) and "config" not in config:
data = {"config": config}
else:
data = config
if "gatk4" in dd.get_tools_on(data):
return _gatk4_cmd(jvm_opts, params, data)
gatk_cmd = utils.which(
os.path.join(os.path.dirname(os.path.realpath(sys.executable)), name))
# if we can't find via the local executable, fallback to being in the path
if not gatk_cmd:
gatk_cmd = utils.which(name)
if gatk_cmd:
return "unset JAVA_HOME && export PATH=%s:$PATH && %s %s %s" % \
(utils.get_java_binpath(gatk_cmd), gatk_cmd,
" ".join(jvm_opts), " ".join([str(x) for x in params]))
def _get_snpeff_cmd(cmd_name, datadir, data, out_file):
"""Retrieve snpEff base command line.
"""
resources = config_utils.get_resources("snpeff", data["config"])
jvm_opts = resources.get("jvm_opts", ["-Xms750m", "-Xmx3g"])
# scale by cores, defaulting to 2x base usage to ensure we have enough memory
# for single core runs to use with human genomes.
# Sets a maximum amount of memory to avoid core dumps exceeding 32Gb
# We shouldn't need that much memory for snpEff, so avoid issues
# https://www.elastic.co/guide/en/elasticsearch/guide/current/heap-sizing.html#compressed_oops
jvm_opts = config_utils.adjust_opts(jvm_opts, {"algorithm": {"memory_adjust":
{"direction": "increase",
"maximum": "30000M",
"magnitude": max(2, dd.get_cores(data))}}})
memory = " ".join(jvm_opts)
snpeff = config_utils.get_program("snpEff", data["config"])
java_args = "-Djava.io.tmpdir=%s" % utils.safe_makedir(os.path.join(os.path.dirname(out_file), "tmp"))
export = "unset JAVA_HOME && export PATH=%s:\"$PATH\" && " % (utils.get_java_binpath())
cmd = "{export} {snpeff} {memory} {java_args} {cmd_name} -dataDir {datadir}"
return cmd.format(**locals())
def run(self, subcmd, opts, memscale=None):
jvm_opts = get_picard_opts(self._config, memscale=memscale)
cmd = ["export", "PATH=%s:\"$PATH\"" % utils.get_java_binpath(), "&&"] + \
[self._cmd] + jvm_opts + [subcmd] + ["%s=%s" % (x, y) for x, y in opts] + \
["VALIDATION_STRINGENCY=SILENT"]
do.run(utils.clear_java_home() + " && " + " ".join(cmd), "Picard: %s" % subcmd)
def run(self, subcmd, opts, memscale=None):
jvm_opts = get_picard_opts(self._config, memscale=memscale)
cmd = ["unset", "JAVA_HOME", "&&", "export", "PATH=%s:$PATH" % utils.get_java_binpath(), "&&"] + \
[self._cmd] + jvm_opts + [subcmd] + ["%s=%s" % (x, y) for x, y in opts] + \
["VALIDATION_STRINGENCY=SILENT"]
do.run(" ".join(cmd), "Picard: %s" % subcmd)
def _gatk4_cmd(jvm_opts, params, data):
"""Retrieve unified command for GATK4, using gatk-launch.
"""
gatk_cmd = utils.which(os.path.join(os.path.dirname(os.path.realpath(sys.executable)), "gatk-launch"))
return "unset JAVA_HOME && export PATH=%s:$PATH && gatk-launch --javaOptions '%s' %s" % \
(utils.get_java_binpath(gatk_cmd), " ".join(jvm_opts), " ".join([str(x) for x in params]))
