def importSingularityImage():
"""Import the Singularity image from Docker if using Singularity."""
mode = os.environ.get("CACTUS_BINARIES_MODE", "docker")
localImage = os.environ.get("CACTUS_USE_LOCAL_SINGULARITY_IMG", "0")
if mode == "singularity":
imgPath = os.environ["CACTUS_SINGULARITY_IMG"]
# If not using local image, pull the docker image
if localImage == "0":
# Singularity will complain if the image file already exists. Remove it.
try:
os.remove(imgPath)
except OSError:
# File doesn't exist
pass
# Singularity 2.4 broke the functionality that let --name
# point to a path instead of a name in the CWD. So we change
# to the proper directory manually, then change back after the
# image is pulled.
# NOTE: singularity writes images in the current directory only
# when SINGULARITY_CACHEDIR is not set
oldCWD = os.getcwd()
os.chdir(os.path.dirname(imgPath))
# --size is deprecated starting in 2.4, but is needed for 2.3 support. Keeping it in for now.
try:
check_call(["singularity", "pull", "--size", "2000", "--name", os.path.basename(imgPath),
"docker://" + getDockerImage()])
except CalledProcessError:
# Call failed, try without --size, required for singularity 3+
check_call(["singularity", "pull", "--name", os.path.basename(imgPath),
"docker://" + getDockerImage()])
os.chdir(oldCWD)
else:
logger.info("Using pre-built singularity image: '{}'".format(imgPath))
def importSingularityImage():
"""Import the Singularity image from Docker if using Singularity."""
mode = os.environ.get("CACTUS_BINARIES_MODE", "docker")
if mode == "singularity":
imgPath = os.environ["CACTUS_SINGULARITY_IMG"]
# Singularity will complain if the image file already exists. Remove it.
try:
os.remove(imgPath)
except OSError:
# File doesn't exist
pass
# Singularity 2.4 broke the functionality that let --name
# point to a path instead of a name in the CWD. So we change
# to the proper directory manually, then change back after the
# image is pulled.
# NOTE: singularity writes images in the current directory only
# when SINGULARITY_CACHEDIR is not set
oldCWD = os.getcwd()
os.chdir(os.path.dirname(imgPath))
# --size is deprecated starting in 2.4, but is needed for 2.3 support. Keeping it in for now.
check_call([
"singularity", "pull", "--size", "2000", "--name",
os.path.basename(imgPath), "docker://" + getDockerImage()
])
os.chdir(oldCWD)
def importSingularityImage(options):
"""Import the Singularity image from Docker if using Singularity."""
mode = os.environ.get("CACTUS_BINARIES_MODE", "docker")
localImage = os.environ.get("CACTUS_USE_LOCAL_SINGULARITY_IMG", "0")
if mode == "singularity" and Toil.parseLocator(
options.jobStore)[0] == "file":
imgPath = os.environ["CACTUS_SINGULARITY_IMG"]
# If not using local image, pull the docker image
if localImage == "0":
# Singularity will complain if the image file already exists. Remove it.
try:
os.remove(imgPath)
except OSError:
# File doesn't exist
pass
# Singularity 2.4 broke the functionality that let --name
# point to a path instead of a name in the CWD. So we change
# to the proper directory manually, then change back after the
# image is pulled.
# NOTE: singularity writes images in the current directory only
# when SINGULARITY_CACHEDIR is not set
oldCWD = os.getcwd()
os.chdir(os.path.dirname(imgPath))
# --size is deprecated starting in 2.4, but is needed for 2.3 support. Keeping it in for now.
try:
check_call([
"singularity", "pull", "--size", "2000", "--name",
os.path.basename(imgPath), "docker://" + getDockerImage()
])
except CalledProcessError:
# Call failed, try without --size, required for singularity 3+
check_call([
"singularity", "pull", "--name",
os.path.basename(imgPath), "docker://" + getDockerImage()
])
os.chdir(oldCWD)
else:
logger.info(
"Using pre-built singularity image: '{}'".format(imgPath))
def main():
parser = ArgumentParser()
parser.add_argument("seqFile", help = "Seq file")
parser.add_argument("--outDir", help='Directory where the processed leaf sequence and ancestral sequences will be placed.'
' Required when not using --wdl')
parser.add_argument("--outSeqFile", help="Path for annotated Seq file output [default: outDir/seqFile]")
parser.add_argument("--outHal", help="Output HAL file [default: outDir/out.hal]")
parser.add_argument("--wdl", action="store_true", help="output wdl workflow instead of list of commands")
parser.add_argument("--noLocalInputs", action="store_true", help="dont embed local input paths in WDL script (as they will need"
" to be respecified when running on Terra")
parser.add_argument("--configFile", default=os.path.join(cactusRootPath(), "cactus_progressive_config.xml"))
parser.add_argument("--preprocessBatchSize", type=int, default=3, help="size (number of genomes) of suggested preprocessing jobs")
parser.add_argument("--jobStore", type=str, default="./jobstore", help="base directory of jobStores to use in suggested commands")
parser.add_argument("--halOptions", type=str, default="--hdf5InMemory", help="options for every hal command")
parser.add_argument("--cactusOptions", type=str, default="--realTimeLogging --logInfo", help="options for every cactus command")
parser.add_argument("--preprocessOnly", action="store_true", help="only decompose into preprocessor and cactus jobs")
parser.add_argument("--dockerImage", type=str, help="docker image to use as wdl runtime")
parser.add_argument("--gpu", action="store_true", help="use gpu-enabled lastz in cactus-blast")
parser.add_argument("--gpuType", default="nvidia-tesla-v100", help="GPU type (to set in WDL runtime parameters)")
parser.add_argument("--gpuCount", default=1, help="GPU count (to set in WDL runtime parameters)")
parser.add_argument("--nvidiaDriver", default="440.64.00", help="Nvidia driver version")
parser.add_argument("--gpuZone", default="us-central1-c", help="zone used for gpu task")
parser.add_argument("--zone", default="us-west2-a", help="zone used for all but gpu tasks")
parser.add_argument("--defaultCores", type=int, help="Number of cores for each job unless otherwise specified")
parser.add_argument("--preprocessCores", type=int, help="Number of cores for each cactus-preprocess job")
parser.add_argument("--blastCores", type=int, help="Number of cores for each cactus-blast job")
parser.add_argument("--alignCores", type=int, help="Number of cores for each cactus-align job")
parser.add_argument("--defaultMem", type=float, help="Memory in GB for each job unless otherwise specified")
parser.add_argument("--preprocessMem", type=float, help="Memory in GB for each cactus-preprocess job")
parser.add_argument("--blastMem", type=float, help="Memory in GB for each cactus-blast job")
parser.add_argument("--alignMem", type=float, help="Memory in GB for each cactus-align job")
parser.add_argument("--defaultDisk", type=int, help="Disk in GB for each job unless otherwise specified")
parser.add_argument("--preprocessDisk", type=int, help="Disk in GB for each cactus-preprocess job")
parser.add_argument("--blastDisk", type=int, help="Disk in GB for each cactus-blast job")
parser.add_argument("--alignDisk", type=int, help="Disk in GB for each cactus-align job")
parser.add_argument("--halAppendDisk", type=int, help="Disk in GB for each halAppendSubtree job")
parser.add_argument("--preprocessPreemptible", type=int, help="Preemptible in GB for each cactus-preprocess job [default=2]", default=2)
parser.add_argument("--blastPreemptible", type=int, help="Preemptible in GB for each cactus-blast job [default=1]", default=1)
parser.add_argument("--alignPreemptible", type=int, help="Preemptible in GB for each cactus-align job [default=1]", default=1)
parser.add_argument("--halAppendPreemptible", type=int, help="Preemptible in GB for each halAppendSubtree job [default=1]", default=1)
options = parser.parse_args()
options.database = 'kyoto_tycoon'
#todo support root option
options.root = None
if not options.wdl:
if not options.outDir:
raise RuntimeError("--outDir option required when not using --wdl")
if not options.outSeqFile:
options.outSeqFile = os.path.join(options.outDir, os.path.basename(options.seqFile))
if os.path.abspath(options.seqFile) == os.path.abspath(options.outSeqFile):
options.outSeqFile += '.1'
if (not options.wdl or not options.gpu) and (options.gpuCount > 1 or options.gpuType != "nvidia-tesla-v100"):
raise RuntimeError("--gpuType and gpuCount can only be used with --wdl --gpu")
if not options.outHal:
options.outHal = os.path.join(options.outDir if options.outDir else '', 'out.hal')
if options.wdl:
if options.preprocessBatchSize != 1:
if options.preprocessBatchSize != 3:
# hacky way to only warn for non-default
sys.stderr.write("Warning: --preprocessBatchSize reset to 1 for --wdl support\n")
options.preprocessBatchSize = 1
# wdl handles output file structure
if options.outDir:
sys.stderr.write("Warning: --outDir option ignored with --wdl\n")
options.outDir = "."
if options.outSeqFile:
sys.stderr.write("Warning: --outSeqFile option ignored with --wdl\n")
options.outSeqFile = None
if options.preprocessOnly:
raise RuntimeError('--preprocessOnly cannot be used in conjunction with --wdl')
if not options.dockerImage:
options.dockerImage = getDockerImage()
# apply defaults
if options.defaultCores:
if not options.preprocessCores:
options.preprocessCores = options.defaultCores
if not options.blastCores:
options.blastCores = options.defaultCores
if not options.alignCores:
options.alignCores = options.defaultCores
if options.defaultMem:
if not options.preprocessMem:
options.preprocessMem = options.defaultMem
if not options.blastMem:
options.blastMem = options.defaultMem
if not options.alignMem:
options.alignMem = options.defaultMem
if not options.alignCores or options.alignCores == 1:
if options.alignCores == 1:
sys.stderr.write("Warning: --alignCores changed from 1 to 2\n")
options.alignCores = 2
if options.defaultDisk:
if not options.preprocessDisk:
options.preprocessDisk = options.defaultDisk
if not options.blastDisk:
options.blastDisk = options.defaultDisk
if not options.alignDisk:
options.alignDisk = options.defaultDisk
if not options.halAppendDisk:
options.halAppendDisk = options.defaultDisk
# https://cromwell.readthedocs.io/en/stable/RuntimeAttributes/#gpucount-gputype-and-nvidiadriverversion
# note: k80 not included as WGA_GPU doesn't run on it.
acceptable_gpus = ['nvidia-tesla-v100', 'nvidia-tesla-p100', 'nvidia-tesla-p4', 'nvidia-tesla-t4']
if options.gpuType not in acceptable_gpus:
raise RuntimeError('--gpuType {} not supported by Terra. Acceptable types are {}'.format(
options.gpuType, acceptable_gpus))
# need to go through this garbage (copied from the main() in progressive_cactus) to
# come up with the project
options.cactusDir = getTempDirectory()
#Create the progressive cactus project
projWrapper = ProjectWrapper(options, options.configFile)
projWrapper.writeXml()
# used to unique jobstore
options.jobStoreCount = 0
pjPath = os.path.join(options.cactusDir, ProjectWrapper.alignmentDirName,
'%s_project.xml' % ProjectWrapper.alignmentDirName)
assert os.path.exists(pjPath)
project = MultiCactusProject()
if not os.path.isdir(options.cactusDir):
os.makedirs(options.cactusDir)
project.readXML(pjPath)
enableDumpStack()
cactusPrepare(options, project)
def main(toil_mode=False):
parser = ArgumentParser()
if toil_mode:
Job.Runner.addToilOptions(parser)
parser.add_argument("--latest", dest="latest", action="store_true",
help="Use the latest version of the docker container "
"rather than pulling one matching this version of cactus")
parser.add_argument("--containerImage", dest="containerImage", default=None,
help="Use the the specified pre-built containter image "
"rather than pulling one from quay.io")
parser.add_argument("--binariesMode", choices=["docker", "local", "singularity"],
help="The way to run the Cactus binaries (at top level; use --cactusOpts to set it in nested calls)",
default=None)
parser.add_argument("seqFile", help = "Seq file")
parser.add_argument("--outDir", help='Directory where the processed leaf sequence and ancestral sequences will be placed.'
' Required when not using --wdl')
parser.add_argument("--outSeqFile", help="Path for annotated Seq file output [default: outDir/seqFile]")
parser.add_argument("--outHal", help="Output HAL file [default: outDir/out.hal]", required=toil_mode)
if not toil_mode:
parser.add_argument("--wdl", action="store_true", help="output wdl workflow instead of list of commands")
parser.add_argument("--noLocalInputs", action="store_true", help="dont embed local input paths in WDL script (as they will need"
" to be respecified when running on Terra")
parser.add_argument("--jobStore", type=str, default="./jobstore", help="base directory of jobStores to use in suggested commands")
parser.add_argument("--configFile", default=os.path.join(cactusRootPath(), "cactus_progressive_config.xml"))
parser.add_argument("--preprocessBatchSize", type=int, default=3, help="size (number of genomes) of suggested preprocessing jobs")
parser.add_argument("--halOptions", type=str, default="--hdf5InMemory", help="options for every hal command")
parser.add_argument("--cactusOptions", type=str, default="--realTimeLogging --logInfo --retryCount 0", help="options for every cactus command")
parser.add_argument("--preprocessOnly", action="store_true", help="only decompose into preprocessor and cactus jobs")
parser.add_argument("--dockerImage", type=str, help="docker image to use as wdl runtime")
parser.add_argument("--gpu", action="store_true", help="use gpu-enabled lastz in cactus-blast")
parser.add_argument("--gpuType", default="nvidia-tesla-v100", help="GPU type (to set in WDL runtime parameters)")
parser.add_argument("--gpuCount", default=1, help="GPU count (to set in WDL runtime parameters)")
parser.add_argument("--nvidiaDriver", default="440.64.00", help="Nvidia driver version")
parser.add_argument("--gpuZone", default="us-central1-c", help="zone used for gpu task")
parser.add_argument("--zone", default="us-west2-a", help="zone used for all but gpu tasks")
if not toil_mode:
parser.add_argument("--defaultCores", type=int, help="Number of cores for each job unless otherwise specified")
parser.add_argument("--preprocessCores", type=int, help="Number of cores for each cactus-preprocess job")
parser.add_argument("--blastCores", type=int, help="Number of cores for each cactus-blast job")
parser.add_argument("--alignCores", type=int, help="Number of cores for each cactus-align job")
if not toil_mode:
parser.add_argument("--defaultMemory", type=human2bytesN, help="Memory for each job unless otherwise specified. "
"Standard suffixes like K, Ki, M, Mi, G or Gi are supported (default=bytes)")
parser.add_argument("--preprocessMemory", type=human2bytesN, help="Memory for each cactus-preprocess job. "
"Standard suffixes like K, Ki, M, Mi, G or Gi are supported (default=bytes)")
parser.add_argument("--blastMemory", type=human2bytesN, help="Memory for each cactus-blast job. "
"Standard suffixes like K, Ki, M, Mi, G or Gi are supported (default=bytes)")
parser.add_argument("--alignMemory", type=human2bytesN, help="Memory for each cactus-align job. "
"Standard suffixes like K, Ki, M, Mi, G or Gi are supported (default=bytes)")
if not toil_mode:
parser.add_argument("--defaultDisk", type=human2bytesN, help="Disk for each job unless otherwise specified. "
"Standard suffixes like K, Ki, M, Mi, G or Gi are supported (default=bytes)")
parser.add_argument("--preprocessDisk", type=human2bytesN, help="Disk for each cactus-preprocess job. "
"Standard suffixes like K, Ki, M, Mi, G or Gi are supported (default=bytes)")
parser.add_argument("--blastDisk", type=human2bytesN, help="Disk for each cactus-blast job. "
"Standard suffixes like K, Ki, M, Mi, G or Gi are supported (default=bytes)")
parser.add_argument("--alignDisk", type=human2bytesN, help="Disk for each cactus-align job. "
"Standard suffixes like K, Ki, M, Mi, G or Gi are supported (default=bytes)")
parser.add_argument("--halAppendDisk", type=human2bytesN, help="Disk for each halAppendSubtree job. "
"Standard suffixes like K, Ki, M, Mi, G or Gi are supported (default=bytes)")
parser.add_argument("--preprocessPreemptible", type=int, help="Preemptible attempt count for each cactus-preprocess job [default=2]", default=2)
parser.add_argument("--blastPreemptible", type=int, help="Preemptible attempt count for each cactus-blast job [default=1]", default=1)
parser.add_argument("--alignPreemptible", type=int, help="Preemptible attempt count for each cactus-align job [default=1]", default=1)
parser.add_argument("--halAppendPreemptible", type=int, help="Preemptible attempt count for each halAppendSubtree job [default=1]", default=1)
parser.add_argument("--database", choices=["kyoto_tycoon", "redis"], help="The type of database", default="kyoto_tycoon")
options = parser.parse_args()
#todo support root option
options.root = None
if toil_mode:
options.wdl = False
options.noLocalInputs = False
options.outDir = '.'
setupBinaries(options)
# need to avoid nested container calls, so set toil-inside-toil jobs to local by default
if "--binariesMode" not in options.cactusOptions:
options.cactusOptions += " --binariesMode local"
if options.jobStore.startswith('aws'):
if not options.outHal.startswith('s3://'):
raise RuntimeError("--outHal must be s3:// address when using s3 job store")
if not has_s3:
raise RuntimeError("S3 support requires toil to be installed with [aws]")
options.toil = toil_mode
if not options.wdl and not options.toil:
if not options.outDir:
raise RuntimeError("--outDir option required when not using --wdl")
if not options.outSeqFile:
options.outSeqFile = os.path.join(options.outDir, os.path.basename(options.seqFile))
if os.path.abspath(options.seqFile) == os.path.abspath(options.outSeqFile):
options.outSeqFile += '.1'
if (not options.wdl or not options.gpu) and (options.gpuCount > 1 or options.gpuType != "nvidia-tesla-v100"):
raise RuntimeError("--gpuType and gpuCount can only be used with --wdl --gpu")
if not options.outHal:
options.outHal = os.path.join(options.outDir if options.outDir else '', 'out.hal')
if options.wdl:
# wdl handles output file structure
if options.outDir:
sys.stderr.write("Warning: --outDir option ignored with --wdl\n")
options.outDir = "."
if options.outSeqFile:
sys.stderr.write("Warning: --outSeqFile option ignored with --wdl\n")
options.outSeqFile = None
if options.preprocessOnly:
raise RuntimeError('--preprocessOnly cannot be used in conjunction with --wdl')
if not options.dockerImage:
options.dockerImage = getDockerImage()
# apply defaults
if options.defaultCores:
if not options.preprocessCores:
options.preprocessCores = options.defaultCores
if not options.blastCores:
options.blastCores = options.defaultCores
if not options.alignCores:
options.alignCores = options.defaultCores
if options.defaultMemory:
if not options.preprocessMemory:
options.preprocessMemory = options.defaultMemory
if not options.blastMemory:
options.blastMemory = options.defaultMemory
if not options.alignMemory:
options.alignMemory = options.defaultMemory
if not options.alignCores or options.alignCores == 1:
if options.alignCores == 1:
sys.stderr.write("Warning: --alignCores changed from 1 to 2\n")
options.alignCores = 2
if options.defaultDisk:
if not options.preprocessDisk:
options.preprocessDisk = options.defaultDisk
if not options.blastDisk:
options.blastDisk = options.defaultDisk
if not options.alignDisk:
options.alignDisk = options.defaultDisk
if not options.halAppendDisk:
options.halAppendDisk = options.defaultDisk
# todo: no reason not to support non-1 batch size, but mirror wdl logic for now
if options.toil:
if options.preprocessBatchSize != 1:
if options.preprocessBatchSize != 3:
# hacky way to only warn for non-default
sys.stderr.write("Warning: --preprocessBatchSize reset to 1 for --wdl support\n")
options.preprocessBatchSize = 1
# todo: could also support this
assert not options.preprocessOnly
# https://cromwell.readthedocs.io/en/stable/RuntimeAttributes/#gpucount-gputype-and-nvidiadriverversion
# note: k80 not included as WGA_GPU doesn't run on it.
acceptable_gpus = ['nvidia-tesla-v100', 'nvidia-tesla-p100', 'nvidia-tesla-p4', 'nvidia-tesla-t4']
if options.gpuType not in acceptable_gpus:
raise RuntimeError('--gpuType {} not supported by Terra. Acceptable types are {}'.format(
options.gpuType, acceptable_gpus))
# need to go through this garbage (copied from the main() in progressive_cactus) to
# come up with the project
options.cactusDir = getTempDirectory()
#Create the progressive cactus project
projWrapper = ProjectWrapper(options, options.configFile)
projWrapper.writeXml()
# used to unique jobstore
options.jobStoreCount = 0
pjPath = os.path.join(options.cactusDir, ProjectWrapper.alignmentDirName,
'%s_project.xml' % ProjectWrapper.alignmentDirName)
assert os.path.exists(pjPath)
project = MultiCactusProject()
if not os.path.isdir(options.cactusDir):
os.makedirs(options.cactusDir)
project.readXML(pjPath)
enableDumpStack()
cactusPrepare(options, project)