def run(self, fileStore):
self.configNode = ET.parse(fileStore.readGlobalFile(self.project.getConfigID())).getroot()
self.configWrapper = ConfigWrapper(self.configNode)
self.configWrapper.substituteAllPredefinedConstantsWithLiterals()
logger.info("Progressive Up: " + self.event)
# open up the experiment
# note that we copy the path into the options here
experimentFile = fileStore.readGlobalFile(self.project.expIDMap[self.event])
expXml = ET.parse(experimentFile).getroot()
experiment = ExperimentWrapper(expXml)
configPath = fileStore.readGlobalFile(experiment.getConfigID())
configXml = ET.parse(configPath).getroot()
seqIDMap = dict()
tree = experiment.getTree()
seqNames = []
for node in tree.postOrderTraversal():
if tree.isLeaf(node):
name = tree.getName(node)
seqIDMap[name] = self.project.outputSequenceIDMap[name]
seqNames.append(name)
logger.info("Sequences in progressive, %s: %s" % (self.event, seqNames))
experimentFile = fileStore.getLocalTempFile()
experiment.writeXML(experimentFile)
self.options.experimentFileID = fileStore.writeGlobalFile(experimentFile)
# take union of command line options and config options for hal and reference
if self.options.buildReference == False:
refNode = findRequiredNode(configXml, "reference")
self.options.buildReference = getOptionalAttrib(refNode, "buildReference", bool, False)
halNode = findRequiredNode(configXml, "hal")
if self.options.buildHal == False:
self.options.buildHal = getOptionalAttrib(halNode, "buildHal", bool, False)
if self.options.buildFasta == False:
self.options.buildFasta = getOptionalAttrib(halNode, "buildFasta", bool, False)
# get parameters that cactus_workflow stuff wants
configFile = fileStore.readGlobalFile(experiment.getConfigID())
configNode = ET.parse(configFile).getroot()
workFlowArgs = CactusWorkflowArguments(self.options, experimentFile=experimentFile, configNode=configNode, seqIDMap = seqIDMap)
# copy over the options so we don't trail them around
workFlowArgs.buildReference = self.options.buildReference
workFlowArgs.buildHal = self.options.buildHal
workFlowArgs.buildFasta = self.options.buildFasta
workFlowArgs.globalLeafEventSet = self.options.globalLeafEventSet
if self.options.intermediateResultsUrl is not None:
# Give the URL prefix a special name for this particular
# subproblem (by suffixing it with the name of the
# internal node in the guide tree)
workFlowArgs.intermediateResultsUrl = self.options.intermediateResultsUrl + '-' + self.event
# Use the trimming strategy to blast ingroups vs outgroups.
finalExpWrapper = self.addChild(CactusTrimmingBlastPhase(cactusWorkflowArguments=workFlowArgs, phaseName="trimBlast")).rv()
logger.info("Going to create alignments and define the cactus tree")
return finalExpWrapper
def progressiveWithSubtreeRootFunction(self, experimentFile, toilDir,
batchSystem, buildAvgs,
buildReference, buildHal,
buildFasta, toilStats):
"""Choose an arbitrary subtree from the larger species tree to run the
alignment on. This function is necessary to keep
runWorkflow_multipleExamples general (specifying a subtree
root doesn't make sense for runCactusWorkflow).
"""
# Get valid internal nodes that are the root of the subtree we
# want to align
expWrapper = ExperimentWrapper(ET.parse(experimentFile).getroot())
tree = expWrapper.getTree()
validNodes = []
for node in tree.postOrderTraversal():
if tree.hasName(node) and not tree.isLeaf(node):
validNodes.append(tree.getName(node))
# Choose a random valid subtree root (NB: the entire species
# tree is a valid subtree)
subtreeRoot = random.choice(validNodes)
logger.info("Chose subtree root %s to test from species tree "
"%s" % (subtreeRoot, NXNewick().writeString(tree)))
self.progressiveFunction(experimentFile, toilDir, batchSystem,
buildAvgs, buildReference, buildHal,
buildFasta, toilStats, subtreeRoot)
def progressiveWithSubtreeRootFunction(self, experimentFile, toilDir,
batchSystem, buildAvgs,
buildReference,
buildHal,
buildFasta,
toilStats):
"""Choose an arbitrary subtree from the larger species tree to run the
alignment on. This function is necessary to keep
runWorkflow_multipleExamples general (specifying a subtree
root doesn't make sense for runCactusWorkflow).
"""
# Get valid internal nodes that are the root of the subtree we
# want to align
expWrapper = ExperimentWrapper(ET.parse(experimentFile).getroot())
tree = expWrapper.getTree()
validNodes = []
for node in tree.postOrderTraversal():
if tree.hasName(node) and not tree.isLeaf(node):
validNodes.append(tree.getName(node))
# Choose a random valid subtree root (NB: the entire species
# tree is a valid subtree)
subtreeRoot = random.choice(validNodes)
logger.info("Chose subtree root %s to test from species tree "
"%s" % (subtreeRoot, NXNewick().writeString(tree)))
self.progressiveFunction(experimentFile, toilDir,
batchSystem, buildAvgs,
buildReference,
buildHal,
buildFasta,
toilStats, subtreeRoot)
def runCreateMultiCactusProject(expFile,
projectFile,
fixNames=False,
outgroupNames=None,
root=None,
overwrite=False):
options = CreateMultiCactusProjectOptions(expFile,
projectFile,
fixNames=fixNames,
outgroupNames=outgroupNames,
root=root,
overwrite=overwrite)
expTemplate = ExperimentWrapper(ET.parse(options.expFile).getroot())
configPath = expTemplate.getConfigPath()
confTemplate = ConfigWrapper(ET.parse(configPath).getroot())
if options.fixNames:
cleanEventTree(expTemplate)
tree = expTemplate.getTree()
if options.outgroupNames is not None:
options.outgroupNames = set(options.outgroupNames)
projNames = set([tree.getName(x) for x in tree.getLeaves()])
for outgroupName in options.outgroupNames:
if outgroupName not in projNames:
raise RuntimeError("Specified outgroup %s not found in tree" %
outgroupName)
mcProj = createMCProject(tree, expTemplate, confTemplate, options)
#Replace the sequences with output sequences
expTemplate.updateTree(mcProj.mcTree, expTemplate.buildSequenceMap())
#Now do the file tree creation
createFileStructure(mcProj, expTemplate, confTemplate, options)
def run(self, fileStore):
self.configNode = ET.parse(fileStore.readGlobalFile(self.project.getConfigID())).getroot()
self.configWrapper = ConfigWrapper(self.configNode)
self.configWrapper.substituteAllPredefinedConstantsWithLiterals()
logger.info("Progressive Up: " + self.event)
# open up the experiment
# note that we copy the path into the options here
experimentFile = fileStore.readGlobalFile(self.project.expIDMap[self.event])
expXml = ET.parse(experimentFile).getroot()
experiment = ExperimentWrapper(expXml)
configPath = fileStore.readGlobalFile(experiment.getConfigID())
configXml = ET.parse(configPath).getroot()
seqIDMap = dict()
tree = experiment.getTree()
seqNames = []
for node in tree.postOrderTraversal():
if tree.isLeaf(node):
name = tree.getName(node)
seqIDMap[name] = self.project.outputSequenceIDMap[name]
seqNames.append(name)
logger.info("Sequences in progressive, %s: %s" % (self.event, seqNames))
experimentFile = fileStore.getLocalTempFile()
experiment.writeXML(experimentFile)
self.options.experimentFileID = fileStore.writeGlobalFile(experimentFile)
# take union of command line options and config options for hal and reference
if self.options.buildReference == False:
refNode = findRequiredNode(configXml, "reference")
self.options.buildReference = getOptionalAttrib(refNode, "buildReference", bool, False)
halNode = findRequiredNode(configXml, "hal")
if self.options.buildHal == False:
self.options.buildHal = getOptionalAttrib(halNode, "buildHal", bool, False)
if self.options.buildFasta == False:
self.options.buildFasta = getOptionalAttrib(halNode, "buildFasta", bool, False)
# get parameters that cactus_workflow stuff wants
configFile = fileStore.readGlobalFile(experiment.getConfigID())
configNode = ET.parse(configFile).getroot()
workFlowArgs = CactusWorkflowArguments(self.options, experimentFile=experimentFile, configNode=configNode, seqIDMap = seqIDMap)
# copy over the options so we don't trail them around
workFlowArgs.buildReference = self.options.buildReference
workFlowArgs.buildHal = self.options.buildHal
workFlowArgs.buildFasta = self.options.buildFasta
workFlowArgs.globalLeafEventSet = self.options.globalLeafEventSet
if self.options.intermediateResultsUrl is not None:
# Give the URL prefix a special name for this particular
# subproblem (by suffixing it with the name of the
# internal node in the guide tree)
workFlowArgs.intermediateResultsUrl = self.options.intermediateResultsUrl + '-' + self.event
# Use the trimming strategy to blast ingroups vs outgroups.
finalExpWrapper = self.addChild(CactusTrimmingBlastPhase(cactusWorkflowArguments=workFlowArgs, phaseName="trimBlast")).rv()
logger.info("Going to create alignments and define the cactus tree")
return finalExpWrapper
def progressiveFunction(self,
experimentFile,
toilDir,
batchSystem,
buildAvgs,
buildHal,
buildFasta,
toilStats,
subtreeRoot=None,
logLevel=None):
eW = ExperimentWrapper(ET.parse(experimentFile).getroot())
seqFile = getTempFile()
with open(seqFile, 'w') as f:
tree = eW.getTree()
newick = NXNewick().writeString(tree)
f.write('%s\n' % newick)
for genome in eW.getGenomesWithSequence():
f.write('%s %s\n' % (genome, eW.getSequenceID(genome)))
config = eW.getConfigPath()
runCactusProgressive(seqFile,
config,
toilDir,
batchSystem=batchSystem,
buildAvgs=buildAvgs,
toilStats=toilStats,
logLevel=logLevel)
def exportHal(job, project, event=None, cacheBytes=None, cacheMDC=None, cacheRDC=None, cacheW0=None, chunk=None, deflate=None, inMemory=True,
checkpointInfo=None):
HALPath = "tmp_alignment.hal"
# traverse tree to make sure we are going breadth-first
tree = project.mcTree
# find subtree if event specified
rootNode = None
if event is not None:
assert event in tree.nameToId and not tree.isLeaf(tree.nameToId[event])
rootNode = tree.nameToId[event]
for node in tree.breadthFirstTraversal(rootNode):
genomeName = tree.getName(node)
if genomeName in project.expMap:
experimentFilePath = job.fileStore.readGlobalFile(project.expIDMap[genomeName])
experiment = ExperimentWrapper(ET.parse(experimentFilePath).getroot())
outgroups = experiment.getOutgroupGenomes()
experiment.setConfigPath(job.fileStore.readGlobalFile(experiment.getConfigID()))
expTreeString = NXNewick().writeString(experiment.getTree(onlyThisSubtree=True))
assert len(expTreeString) > 1
assert experiment.getHalID() is not None
assert experiment.getHalFastaID() is not None
subHALPath = job.fileStore.readGlobalFile(experiment.getHalID())
halFastaPath = job.fileStore.readGlobalFile(experiment.getHalFastaID())
args = [os.path.basename(subHALPath), os.path.basename(halFastaPath), expTreeString, os.path.basename(HALPath)]
if len(outgroups) > 0:
args += ["--outgroups", ",".join(outgroups)]
if cacheBytes is not None:
args += ["--cacheBytes", cacheBytes]
if cacheMDC is not None:
args += ["--cacheMDC", cacheMDC]
if cacheRDC is not None:
args += ["--cacheRDC", cacheRDC]
if cacheW0 is not None:
args += ["--cacheW0", cacheW0]
if chunk is not None:
args += ["--chunk", chunk]
if deflate is not None:
args += ["--deflate", deflate]
if inMemory is True:
args += ["--inMemory"]
cactus_call(parameters=["halAppendCactusSubtree"] + args)
cactus_call(parameters=["halSetMetadata", HALPath, "CACTUS_COMMIT", cactus_commit])
with job.fileStore.readGlobalFileStream(project.configID) as configFile:
cactus_call(parameters=["halSetMetadata", HALPath, "CACTUS_CONFIG", b64encode(configFile.read()).decode()])
if checkpointInfo:
write_s3(HALPath, checkpointInfo[1], region=checkpointInfo[0])
return job.fileStore.writeGlobalFile(HALPath)
def testSequenceMap(self):
xmlRoot = self.__makeXmlDummy(self.tree, self.sequences)
exp = ExperimentWrapper(xmlRoot)
assert NXNewick().writeString(exp.getTree()) == self.tree
seqMap = exp.buildSequenceMap()
seqList = self.sequences.split()
for i in seqList:
assert seqMap[os.path.splitext(i)[0].upper()] == i
def testSequenceMap(self):
xmlRoot = self.__makeXmlDummy(self.tree, self.sequences)
exp = ExperimentWrapper(xmlRoot)
assert NXNewick().writeString(exp.getTree()) == self.tree
seqMap = exp.buildSequenceMap()
seqList = self.sequences.split()
for i in seqList:
assert seqMap[os.path.splitext(i)[0].upper()] == i
def main():
usage = "usage: %prog [options] <experiment> <output project path>"
description = "Setup a multi-cactus project using an experiment xml as template"
parser = OptionParser(usage=usage, description=description)
parser.add_option("--fixNames", dest="fixNames", default = "True",
help="try to make sequence and event names MAF-compliant [default=true]")
parser.add_option("--outgroupNames", dest="outgroupNames", default = None,
help="comma-separated names of high quality assemblies to use as outgroups [default=everything]")
parser.add_option("--root", dest="root", type=str,
help="name of alignment root (must be labeled ancestral node in tree in input experiment). Useful "
"for allowing the tree to contain nodes that won't be in the alignment but can still be used for "
"outgroups.",
default=None)
parser.add_option("--overwrite", action="store_true", help="Overwrite existing experiment files", default=False)
options, args = parser.parse_args()
if len(args) != 2:
parser.print_help()
raise RuntimeError("Wrong number of arguments")
options.expFile = args[0]
options.path = os.path.abspath(args[1])
options.name = os.path.basename(options.path)
options.fixNames = not options.fixNames.lower() == "false"
if (os.path.isdir(options.path) and not options.overwrite) or os.path.isfile(options.path):
raise RuntimeError("Output project path %s exists\n" % options.path)
expTemplate = ExperimentWrapper(ET.parse(options.expFile).getroot())
configPath = expTemplate.getConfigPath()
confTemplate = ConfigWrapper(ET.parse(configPath).getroot())
if options.fixNames:
cleanEventTree(expTemplate)
checkInputSequencePaths(expTemplate)
tree = expTemplate.getTree()
# Check that the tree is sensible (root has at least 1 child)
if len(tree.getChildren(tree.getRootId())) == 0:
raise RuntimeError("Input species tree has only one node.")
if options.outgroupNames is not None:
projNames = set([tree.getName(x) for x in tree.getLeaves()])
options.outgroupNames = set(options.outgroupNames.split(","))
for outgroupName in options.outgroupNames:
if outgroupName not in projNames:
raise RuntimeError("Specified outgroup %s not found in tree" % outgroupName)
mcProj = createMCProject(tree, expTemplate, confTemplate, options)
#Replace the sequences with output sequences
expTemplate.updateTree(mcProj.mcTree, expTemplate.buildSequenceMap())
expTemplate.setSequences(CactusPreprocessor.getOutputSequenceFiles(mcProj.inputSequences, expTemplate.getOutputSequenceDir()))
#Now do the file tree creation
createFileStructure(mcProj, expTemplate, confTemplate, options)
# mcProj.check()
return 0
def progressiveFunction(self, experimentFile, toilDir,
batchSystem, buildAvgs,
buildHal,
buildFasta,
toilStats,
subtreeRoot=None):
eW = ExperimentWrapper(ET.parse(experimentFile).getroot())
seqFile = getTempFile()
with open(seqFile, 'w') as f:
tree = eW.getTree()
def get_leaves_and_outgroups(options, project, root):
""" fish the leaves and outgroups out of the experiment xml """
# open up the experiment (as we do in ProgressiveUp.run)
experimentFile = project.expMap[root]
expXml = ET.parse(experimentFile).getroot()
experiment = ExperimentWrapper(expXml)
tree = MultiCactusTree(experiment.getTree()).extractSubTree(root)
leaves = tree.getChildNames(tree.getRootName())
outgroups = experiment.getOutgroupGenomes()
return leaves, outgroups
def exportHal(job, project, event=None, cacheBytes=None, cacheMDC=None, cacheRDC=None, cacheW0=None, chunk=None, deflate=None, inMemory=False):
HALPath = "tmp_alignment.hal"
# traverse tree to make sure we are going breadth-first
tree = project.mcTree
# find subtree if event specified
rootNode = None
if event is not None:
assert event in tree.nameToId and not tree.isLeaf(tree.nameToId[event])
rootNode = tree.nameToId[event]
for node in tree.breadthFirstTraversal(rootNode):
genomeName = tree.getName(node)
if genomeName in project.expMap:
experimentFilePath = job.fileStore.readGlobalFile(project.expIDMap[genomeName])
experiment = ExperimentWrapper(ET.parse(experimentFilePath).getroot())
outgroups = experiment.getOutgroupEvents()
experiment.setConfigPath(job.fileStore.readGlobalFile(experiment.getConfigID()))
expTreeString = NXNewick().writeString(experiment.getTree(onlyThisSubtree=True))
assert len(expTreeString) > 1
assert experiment.getHalID() is not None
assert experiment.getHalFastaID() is not None
subHALPath = job.fileStore.readGlobalFile(experiment.getHalID())
halFastaPath = job.fileStore.readGlobalFile(experiment.getHalFastaID())
args = [os.path.basename(subHALPath), os.path.basename(halFastaPath), expTreeString, os.path.basename(HALPath)]
if len(outgroups) > 0:
args += ["--outgroups", ",".join(outgroups)]
if cacheBytes is not None:
args += ["--cacheBytes", cacheBytes]
if cacheMDC is not None:
args += ["--cacheMDC", cacheMDC]
if cacheRDC is not None:
args += ["--cacheRDC", cacheRDC]
if cacheW0 is not None:
args += ["--cacheW0", cacheW0]
if chunk is not None:
args += ["--chunk", chunk]
if deflate is not None:
args += ["--deflate", deflate]
if inMemory is True:
args += ["--inMemory"]
cactus_call(parameters=["halAppendCactusSubtree"] + args)
cactus_call(parameters=["halSetMetadata", HALPath, "CACTUS_COMMIT", cactus_commit])
with job.fileStore.readGlobalFileStream(project.configID) as configFile:
cactus_call(parameters=["halSetMetadata", HALPath, "CACTUS_CONFIG", b64encode(configFile.read())])
return job.fileStore.writeGlobalFile(HALPath)
def testOutgroups(self):
xmlRoot = self.__makeXmlDummy(self.tree, self.sequences)
exp = ExperimentWrapper(xmlRoot)
assert NXNewick().writeString(exp.getTree()) == self.tree
exp.addOutgroupSequence("outgroup", 1.3, "outgroup.fa")
exp.addOutgroupSequence("outgroup2", 2.6, "outgroup2.fa")
assert exp.getOutgroupEvents() == ["outgroup", "outgroup2"]
seqMap = exp.buildSequenceMap()
assert "outgroup" in seqMap
assert seqMap["outgroup"] == "outgroup.fa"
assert "outgroup2" in seqMap
assert seqMap["outgroup2"] == "outgroup2.fa"
def loadProject(self, mcProject, fileStore = None):
self.inGraph = NX.DiGraph()
globTree = mcProject.mcTree
self.maxParallelSubtrees = None
leafEvents = [globTree.getName(i) for i in globTree.getLeaves()]
expMap = None
if fileStore:
expMap = dict()
for name in mcProject.expIDMap:
expMap[name] = fileStore.readGlobalFile(mcProject.expIDMap[name])
else:
expMap = mcProject.expMap
for name, expPath in list(expMap.items()):
exp = ExperimentWrapper(ET.parse(expPath).getroot())
tree = exp.getTree()
self.inGraph.add_node(name)
# Go through the species tree and add the correct
# dependencies (i.e. to the outgroups and the ingroups,
# but not to the other nodes that are just there because
# they are needed to form the correct paths).
for node in tree.postOrderTraversal():
nodeName = tree.getName(node)
# we don't add edges for leaves (in the global tree)
# as they are input sequences and do not form dependencies
# (it would be clever to maybe do the same with existing
# references when --overwrite is not specified but for now
# we just do the leaves)
if nodeName not in leafEvents and tree.isLeaf(node):
self.inGraph.add_edge(name, nodeName)
if fileStore:
configFile = fileStore.readGlobalFile(exp.getConfigID())
else:
# hack from running from cactus-prepare
configFile = exp.getConfigPath()
configElem = ET.parse(configFile).getroot()
conf = ConfigWrapper(configElem)
# load max parellel subtrees from the node's config
if self.maxParallelSubtrees is None:
self.maxParallelSubtrees = conf.getMaxParallelSubtrees()
else:
assert self.maxParallelSubtrees == conf.getMaxParallelSubtrees()
assert NX.is_directed_acyclic_graph(self.inGraph)
def loadProject(self, mcProject):
self.inGraph = NX.DiGraph()
globTree = mcProject.mcTree
self.maxParallelSubtrees = None
leafEvents = [globTree.getName(i) for i in globTree.getLeaves()]
for name, expPath in mcProject.expMap.items():
exp = ExperimentWrapper(ET.parse(expPath).getroot())
tree = exp.getTree()
self.inGraph.add_node(name)
# Go through the species tree and add the correct
# dependencies (i.e. to the outgroups and the ingroups,
# but not to the other nodes that are just there because
# they are needed to form the correct paths).
for node in tree.postOrderTraversal():
nodeName = tree.getName(node)
if not tree.isLeaf(node) and nodeName not in exp.getOutgroupEvents():
# This node is just an internal node added while
# creating the induced tree from the species
# tree. None of the sequence is used, so skip it.
continue
assert tree.hasParent(node)
if nodeName not in exp.getOutgroupEvents() and tree.getName(tree.getParent(node)) != name:
# This leaf isn't an ingroup or an outgroup, it was
# just added to make the species tree
# binary. (Hopefully this will be unnecessary in
# the future.)
continue
# we don't add edges for leaves (in the global tree)
# as they are input sequences and do not form dependencies
# (it would be clever to maybe do the same with existing
# references when --overwrite is not specified but for now
# we just do the leaves)
if nodeName not in leafEvents:
self.inGraph.add_edge(name, nodeName)
configElem = ET.parse(exp.getConfig()).getroot()
conf = ConfigWrapper(configElem)
# load max parellel subtrees from the node's config
if self.maxParallelSubtrees is None:
self.maxParallelSubtrees = conf.getMaxParallelSubtrees()
else:
assert self.maxParallelSubtrees == conf.getMaxParallelSubtrees()
assert NX.is_directed_acyclic_graph(self.inGraph)
def loadProject(self, mcProject, fileStore = None):
self.inGraph = NX.DiGraph()
globTree = mcProject.mcTree
self.maxParallelSubtrees = None
leafEvents = [globTree.getName(i) for i in globTree.getLeaves()]
expMap = None
if fileStore:
expMap = dict()
for name in mcProject.expIDMap:
expMap[name] = fileStore.readGlobalFile(mcProject.expIDMap[name])
else:
expMap = mcProject.expMap
for name, expPath in expMap.items():
exp = ExperimentWrapper(ET.parse(expPath).getroot())
tree = exp.getTree()
self.inGraph.add_node(name)
# Go through the species tree and add the correct
# dependencies (i.e. to the outgroups and the ingroups,
# but not to the other nodes that are just there because
# they are needed to form the correct paths).
for node in tree.postOrderTraversal():
nodeName = tree.getName(node)
# we don't add edges for leaves (in the global tree)
# as they are input sequences and do not form dependencies
# (it would be clever to maybe do the same with existing
# references when --overwrite is not specified but for now
# we just do the leaves)
if nodeName not in leafEvents and nodeName in exp.getSequenceMap():
self.inGraph.add_edge(name, nodeName)
configFile = fileStore.readGlobalFile(exp.getConfigID())
configElem = ET.parse(configFile).getroot()
conf = ConfigWrapper(configElem)
# load max parellel subtrees from the node's config
if self.maxParallelSubtrees is None:
self.maxParallelSubtrees = conf.getMaxParallelSubtrees()
else:
assert self.maxParallelSubtrees == conf.getMaxParallelSubtrees()
assert NX.is_directed_acyclic_graph(self.inGraph)
def progressiveWithSubtreeRootFunction(self,
experimentFile,
toilDir,
batchSystem,
buildAvgs,
buildHal,
buildFasta,
toilStats,
logLevel=None):
"""Choose an arbitrary subtree from the larger species tree to run the
alignment on. This function is necessary to keep
runWorkflow_multipleExamples general (specifying a subtree
root doesn't make sense for runCactusWorkflow).
"""
# Get valid internal nodes that are the root of the subtree we
# want to align
expWrapper = ExperimentWrapper(ET.parse(experimentFile).getroot())
tree = expWrapper.getTree()
validNodes = []
for node in tree.postOrderTraversal():
if tree.hasName(node) and not tree.isLeaf(node) and tree.hasParent(
node):
validNodes.append(tree.getName(node))
# Choose a random valid subtree root (excluding the species tree root)
subtreeRoot = random.choice(validNodes)
self.progressiveFunction(experimentFile,
toilDir,
batchSystem,
buildAvgs,
buildHal,
buildFasta,
toilStats,
subtreeRoot,
logLevel=logLevel)
def runCactusAfterBlastOnly(options):
with Toil(options) as toil:
importSingularityImage(options)
#Run the workflow
if options.restart:
halID = toil.restart()
else:
options.cactusDir = getTempDirectory()
# apply path overrides. this was necessary for wdl which doesn't take kindly to
# text files of local paths (ie seqfile). one way to fix would be to add support
# for s3 paths and force wdl to use it. a better way would be a more fundamental
# interface shift away from files of paths throughout all of cactus
if options.pathOverrides:
seqFile = SeqFile(options.seqFile)
configNode = ET.parse(options.configFile).getroot()
config = ConfigWrapper(configNode)
tree = MultiCactusTree(seqFile.tree)
tree.nameUnlabeledInternalNodes(
prefix=config.getDefaultInternalNodePrefix())
for name, override in zip(options.pathOverrideNames,
options.pathOverrides):
seqFile.pathMap[name] = override
override_seq = os.path.join(options.cactusDir,
'seqFile.override')
with open(override_seq, 'w') as out_sf:
out_sf.write(str(seqFile))
options.seqFile = override_seq
#to be consistent with all-in-one cactus, we make sure the project
#isn't limiting itself to the subtree (todo: parameterize so root can
#be passed through from prepare to blast/align)
proj_options = copy.deepcopy(options)
proj_options.root = None
#Create the progressive cactus project (as we do in runCactusProgressive)
projWrapper = ProjectWrapper(proj_options,
proj_options.configFile,
ignoreSeqPaths=options.root)
projWrapper.writeXml()
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)
# open up the experiment (as we do in ProgressiveUp.run)
# note that we copy the path into the options here
experimentFile = project.expMap[options.root]
expXml = ET.parse(experimentFile).getroot()
experiment = ExperimentWrapper(expXml)
configPath = experiment.getConfigPath()
configXml = ET.parse(configPath).getroot()
seqIDMap = dict()
tree = MultiCactusTree(experiment.getTree()).extractSubTree(
options.root)
leaves = [tree.getName(leaf) for leaf in tree.getLeaves()]
outgroups = experiment.getOutgroupGenomes()
genome_set = set(leaves + outgroups)
# this is a hack to allow specifying all the input on the command line, rather than using suffix lookups
def get_input_path(suffix=''):
base_path = options.cigarsFile[0]
for input_path in options.cigarsFile:
if suffix and input_path.endswith(suffix):
return input_path
if os.path.basename(base_path).startswith(
os.path.basename(input_path)):
base_path = input_path
return base_path + suffix
# import the outgroups
outgroupIDs = []
outgroup_fragment_found = False
for i, outgroup in enumerate(outgroups):
try:
outgroupID = toil.importFile(
makeURL(get_input_path('.og_fragment_{}'.format(i))))
outgroupIDs.append(outgroupID)
experiment.setSequenceID(outgroup, outgroupID)
outgroup_fragment_found = True
assert not options.pangenome
except:
# we assume that input is not coming from cactus blast, so we'll treat output
# sequences normally and not go looking for fragments
outgroupIDs = []
break
#import the sequences (that we need to align for the given event, ie leaves and outgroups)
for genome, seq in list(project.inputSequenceMap.items()):
if genome in leaves or (not outgroup_fragment_found
and genome in outgroups):
if os.path.isdir(seq):
tmpSeq = getTempFile()
catFiles([
os.path.join(seq, subSeq)
for subSeq in os.listdir(seq)
], tmpSeq)
seq = tmpSeq
seq = makeURL(seq)
experiment.setSequenceID(genome, toil.importFile(seq))
if not outgroup_fragment_found:
outgroupIDs = [
experiment.getSequenceID(outgroup)
for outgroup in outgroups
]
# write back the experiment, as CactusWorkflowArguments wants a path
experiment.writeXML(experimentFile)
#import cactus config
if options.configFile:
cactusConfigID = toil.importFile(makeURL(options.configFile))
else:
cactusConfigID = toil.importFile(
makeURL(project.getConfigPath()))
project.setConfigID(cactusConfigID)
project.syncToFileStore(toil)
configNode = ET.parse(project.getConfigPath()).getroot()
configWrapper = ConfigWrapper(configNode)
configWrapper.substituteAllPredefinedConstantsWithLiterals()
if options.pangenome:
# turn off the megablock filter as it ruins non-all-to-all alignments
configWrapper.disableCafMegablockFilter()
# the recoverable chains parameter does not seem to play nicely with star-like alignments either
#configWrapper.disableRecoverableChains()
workFlowArgs = CactusWorkflowArguments(
options,
experimentFile=experimentFile,
configNode=configNode,
seqIDMap=project.inputSequenceIDMap)
#import the files that cactus-blast made
workFlowArgs.alignmentsID = toil.importFile(
makeURL(get_input_path()))
workFlowArgs.secondaryAlignmentsID = None
if not options.pafInput:
try:
workFlowArgs.secondaryAlignmentsID = toil.importFile(
makeURL(get_input_path('.secondary')))
except:
pass
workFlowArgs.outgroupFragmentIDs = outgroupIDs
workFlowArgs.ingroupCoverageIDs = []
if outgroup_fragment_found and len(outgroups) > 0:
for i in range(len(leaves)):
workFlowArgs.ingroupCoverageIDs.append(
toil.importFile(
makeURL(get_input_path(
'.ig_coverage_{}'.format(i)))))
halID = toil.start(
Job.wrapJobFn(run_cactus_align,
configWrapper,
workFlowArgs,
project,
doRenaming=options.nonCactusInput,
pafInput=options.pafInput))
# export the hal
toil.exportFile(halID, makeURL(options.outputHal))
def main():
args = initParser()
myProj = MultiCactusProject()
myProj.readXML(args['cactus_project'])
if not args['append']:
# Overwrite existing hal
print 'rm -f {0}'.format(args['HAL_file_path'])
system('rm -f {0}'.format(args['HAL_file_path']))
# some quick stats
totalTime = time.time()
totalAppendTime = 0
# traverse tree to make sure we are going breadth-first
tree = myProj.mcTree
# find subtree if event specified
event = args['event']
rootNode = None
if event is not None:
assert event in tree.nameToId and not tree.isLeaf(tree.nameToId[event])
rootNode = tree.nameToId[event]
for node in tree.breadthFirstTraversal(rootNode):
genomeName = tree.getName(node)
if genomeName in myProj.expMap:
experimentFilePath = myProj.expMap[genomeName]
experiment = ExperimentWrapper(
ET.parse(experimentFilePath).getroot())
outgroups = experiment.getOutgroupEvents()
expTreeString = NXNewick().writeString(experiment.getTree())
assert len(expTreeString) > 1
assert experiment.getHALPath() is not None
assert experiment.getHALFastaPath() is not None
cmdline = "time halAppendCactusSubtree \'{0}\' \'{1}\' \'{2}\' \'{3}\'".format(
experiment.getHALPath(), experiment.getHALFastaPath(),
expTreeString, args['HAL_file_path'])
if len(outgroups) > 0:
cmdline += " --outgroups {0}".format(",".join(outgroups))
if args["cacheBytes"] is not None:
cmdline += " --cacheBytes {0}".format(args["cacheBytes"])
if args["cacheMDC"] is not None:
cmdline += " --cacheMDC {0}".format(args["cacheMDC"])
if args["cacheRDC"] is not None:
cmdline += " --cacheRDC {0}".format(args["cacheRDC"])
if args["cacheW0"] is not None:
cmdline += " --cacheW0 {0}".format(args["cacheW0"])
if args["chunk"] is not None:
cmdline += " --chunk {0}".format(args["chunk"])
if args["deflate"] is not None:
cmdline += " --deflate {0}".format(args["deflate"])
if args["inMemory"] is True:
cmdline += " --inMemory"
print cmdline
appendTime = time.time()
system(cmdline)
appendTime = time.time() - appendTime
totalAppendTime += appendTime
# print "time of above command: {0:.2f}".format(appendTime)
totalTime = time.time() - totalTime
print "total time: {0:.2f} total halAppendCactusSubtree time: {1:.2f}".format(
totalTime, totalAppendTime)
def make_align_job(options, toil):
options.cactusDir = getTempDirectory()
# apply path overrides. this was necessary for wdl which doesn't take kindly to
# text files of local paths (ie seqfile). one way to fix would be to add support
# for s3 paths and force wdl to use it. a better way would be a more fundamental
# interface shift away from files of paths throughout all of cactus
if options.pathOverrides:
seqFile = SeqFile(options.seqFile)
configNode = ET.parse(options.configFile).getroot()
config = ConfigWrapper(configNode)
tree = MultiCactusTree(seqFile.tree)
tree.nameUnlabeledInternalNodes(
prefix=config.getDefaultInternalNodePrefix())
for name, override in zip(options.pathOverrideNames,
options.pathOverrides):
seqFile.pathMap[name] = override
override_seq = os.path.join(options.cactusDir, 'seqFile.override')
with open(override_seq, 'w') as out_sf:
out_sf.write(str(seqFile))
options.seqFile = override_seq
if not options.root:
seqFile = SeqFile(options.seqFile)
configNode = ET.parse(options.configFile).getroot()
config = ConfigWrapper(configNode)
mcTree = MultiCactusTree(seqFile.tree)
mcTree.nameUnlabeledInternalNodes(
prefix=config.getDefaultInternalNodePrefix())
options.root = mcTree.getRootName()
if options.acyclic:
seqFile = SeqFile(options.seqFile)
tree = MultiCactusTree(seqFile.tree)
leaves = [tree.getName(leaf) for leaf in tree.getLeaves()]
if options.acyclic not in leaves:
raise RuntimeError(
"Genome specified with --acyclic, {}, not found in tree leaves"
.format(options.acyclic))
#to be consistent with all-in-one cactus, we make sure the project
#isn't limiting itself to the subtree (todo: parameterize so root can
#be passed through from prepare to blast/align)
proj_options = copy.deepcopy(options)
proj_options.root = None
#Create the progressive cactus project (as we do in runCactusProgressive)
projWrapper = ProjectWrapper(proj_options,
proj_options.configFile,
ignoreSeqPaths=options.root)
projWrapper.writeXml()
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)
# open up the experiment (as we do in ProgressiveUp.run)
# note that we copy the path into the options here
experimentFile = project.expMap[options.root]
expXml = ET.parse(experimentFile).getroot()
experiment = ExperimentWrapper(expXml)
configPath = experiment.getConfigPath()
configXml = ET.parse(configPath).getroot()
seqIDMap = dict()
tree = MultiCactusTree(experiment.getTree()).extractSubTree(options.root)
leaves = [tree.getName(leaf) for leaf in tree.getLeaves()]
outgroups = experiment.getOutgroupGenomes()
genome_set = set(leaves + outgroups)
# this is a hack to allow specifying all the input on the command line, rather than using suffix lookups
def get_input_path(suffix=''):
base_path = options.cigarsFile[0]
for input_path in options.cigarsFile:
if suffix and input_path.endswith(suffix):
return input_path
if os.path.basename(base_path).startswith(
os.path.basename(input_path)):
base_path = input_path
return base_path + suffix
# import the outgroups
outgroupIDs = []
outgroup_fragment_found = False
for i, outgroup in enumerate(outgroups):
try:
outgroupID = toil.importFile(
makeURL(get_input_path('.og_fragment_{}'.format(i))))
outgroupIDs.append(outgroupID)
experiment.setSequenceID(outgroup, outgroupID)
outgroup_fragment_found = True
assert not options.pangenome
except:
# we assume that input is not coming from cactus blast, so we'll treat output
# sequences normally and not go looking for fragments
outgroupIDs = []
break
#import the sequences (that we need to align for the given event, ie leaves and outgroups)
for genome, seq in list(project.inputSequenceMap.items()):
if genome in leaves or (not outgroup_fragment_found
and genome in outgroups):
if os.path.isdir(seq):
tmpSeq = getTempFile()
catFiles(
[os.path.join(seq, subSeq) for subSeq in os.listdir(seq)],
tmpSeq)
seq = tmpSeq
seq = makeURL(seq)
logger.info("Importing {}".format(seq))
experiment.setSequenceID(genome, toil.importFile(seq))
if not outgroup_fragment_found:
outgroupIDs = [
experiment.getSequenceID(outgroup) for outgroup in outgroups
]
# write back the experiment, as CactusWorkflowArguments wants a path
experiment.writeXML(experimentFile)
#import cactus config
if options.configFile:
cactusConfigID = toil.importFile(makeURL(options.configFile))
else:
cactusConfigID = toil.importFile(makeURL(project.getConfigPath()))
project.setConfigID(cactusConfigID)
project.syncToFileStore(toil)
configNode = ET.parse(project.getConfigPath()).getroot()
configWrapper = ConfigWrapper(configNode)
configWrapper.substituteAllPredefinedConstantsWithLiterals()
if options.singleCopySpecies:
findRequiredNode(
configWrapper.xmlRoot,
"caf").attrib["alignmentFilter"] = "singleCopyEvent:{}".format(
options.singleCopySpecies)
if options.barMaskFilter:
findRequiredNode(
configWrapper.xmlRoot,
"bar").attrib["partialOrderAlignmentMaskFilter"] = str(
options.barMaskFilter)
if options.pangenome:
# turn off the megablock filter as it ruins non-all-to-all alignments
findRequiredNode(configWrapper.xmlRoot,
"caf").attrib["minimumBlockHomologySupport"] = "0"
findRequiredNode(
configWrapper.xmlRoot,
"caf").attrib["minimumBlockDegreeToCheckSupport"] = "9999999999"
# turn off mapq filtering
findRequiredNode(configWrapper.xmlRoot,
"caf").attrib["runMapQFiltering"] = "0"
# more iterations here helps quite a bit to reduce underalignment
findRequiredNode(configWrapper.xmlRoot,
"caf").attrib["maxRecoverableChainsIterations"] = "50"
# turn down minimum block degree to get a fat ancestor
findRequiredNode(configWrapper.xmlRoot,
"bar").attrib["minimumBlockDegree"] = "1"
# turn on POA
findRequiredNode(configWrapper.xmlRoot,
"bar").attrib["partialOrderAlignment"] = "1"
# save it
if not options.batch:
pg_file = options.outHal + ".pg-conf.xml"
if pg_file.startswith('s3://'):
pg_temp_file = getTempFile()
else:
pg_temp_file = pg_file
configWrapper.writeXML(pg_temp_file)
if pg_file.startswith('s3://'):
write_s3(pg_temp_file,
pg_file,
region=get_aws_region(options.jobStore))
logger.info("pangenome configuration overrides saved in {}".format(
pg_file))
workFlowArgs = CactusWorkflowArguments(options,
experimentFile=experimentFile,
configNode=configNode,
seqIDMap=project.inputSequenceIDMap)
#import the files that cactus-blast made
workFlowArgs.alignmentsID = toil.importFile(makeURL(get_input_path()))
workFlowArgs.secondaryAlignmentsID = None
if not options.pafInput:
try:
workFlowArgs.secondaryAlignmentsID = toil.importFile(
makeURL(get_input_path('.secondary')))
except:
pass
workFlowArgs.outgroupFragmentIDs = outgroupIDs
workFlowArgs.ingroupCoverageIDs = []
if outgroup_fragment_found and len(outgroups) > 0:
for i in range(len(leaves)):
workFlowArgs.ingroupCoverageIDs.append(
toil.importFile(
makeURL(get_input_path('.ig_coverage_{}'.format(i)))))
align_job = Job.wrapJobFn(run_cactus_align,
configWrapper,
workFlowArgs,
project,
checkpointInfo=options.checkpointInfo,
doRenaming=options.nonCactusInput,
pafInput=options.pafInput,
pafSecondaries=options.usePafSecondaries,
doVG=options.outVG,
doGFA=options.outGFA,
delay=options.stagger,
eventNameAsID=options.eventNameAsID,
acyclicEvent=options.acyclic)
return align_job
def runCactusAfterBlastOnly(options):
with Toil(options) as toil:
importSingularityImage(options)
#Run the workflow
if options.restart:
alignmentID = toil.restart()
else:
options.cactusDir = getTempDirectory()
#Create the progressive cactus project (as we do in runCactusProgressive)
projWrapper = ProjectWrapper(options,
options.configFile,
ignoreSeqPaths=options.root)
projWrapper.writeXml()
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)
# open up the experiment (as we do in ProgressiveUp.run)
# note that we copy the path into the options here
experimentFile = project.expMap[options.root]
expXml = ET.parse(experimentFile).getroot()
experiment = ExperimentWrapper(expXml)
configPath = experiment.getConfigPath()
configXml = ET.parse(configPath).getroot()
seqIDMap = dict()
tree = MultiCactusTree(experiment.getTree()).extractSubTree(
options.root)
leaves = [tree.getName(leaf) for leaf in tree.getLeaves()]
outgroups = experiment.getOutgroupGenomes()
genome_set = set(leaves + outgroups)
# import the outgroups
outgroupIDs = []
cactus_blast_input = not options.nonBlastInput
for i, outgroup in enumerate(outgroups):
try:
outgroupID = toil.importFile(
makeURL(options.blastOutput) +
'.og_fragment_{}'.format(i))
outgroupIDs.append(outgroupID)
experiment.setSequenceID(outgroup, outgroupID)
except:
if cactus_blast_input:
raise
# we assume that input is not coming from cactus blast, so we'll treat output
# sequences normally and not go looking for fragments
outgroupIDs = []
break
#import the sequences (that we need to align for the given event, ie leaves and outgroups)
for genome, seq in list(project.inputSequenceMap.items()):
if genome in leaves or (not cactus_blast_input
and genome in outgroups):
if os.path.isdir(seq):
tmpSeq = getTempFile()
catFiles([
os.path.join(seq, subSeq)
for subSeq in os.listdir(seq)
], tmpSeq)
seq = tmpSeq
seq = makeURL(seq)
experiment.setSequenceID(genome, toil.importFile(seq))
if not cactus_blast_input:
outgroupIDs = [
experiment.getSequenceID(outgroup)
for outgroup in outgroups
]
# write back the experiment, as CactusWorkflowArguments wants a path
experiment.writeXML(experimentFile)
#import cactus config
if options.configFile:
cactusConfigID = toil.importFile(makeURL(options.configFile))
else:
cactusConfigID = toil.importFile(
makeURL(project.getConfigPath()))
project.setConfigID(cactusConfigID)
project.syncToFileStore(toil)
configNode = ET.parse(project.getConfigPath()).getroot()
configWrapper = ConfigWrapper(configNode)
configWrapper.substituteAllPredefinedConstantsWithLiterals()
workFlowArgs = CactusWorkflowArguments(
options,
experimentFile=experimentFile,
configNode=configNode,
seqIDMap=project.inputSequenceIDMap)
#import the files that cactus-blast made
workFlowArgs.alignmentsID = toil.importFile(
makeURL(options.blastOutput))
try:
workFlowArgs.secondaryAlignmentsID = toil.importFile(
makeURL(options.blastOutput) + '.secondary')
except:
workFlowArgs.secondaryAlignmentsID = None
workFlowArgs.outgroupFragmentIDs = outgroupIDs
workFlowArgs.ingroupCoverageIDs = []
if cactus_blast_input and len(outgroups) > 0:
for i in range(len(leaves)):
workFlowArgs.ingroupCoverageIDs.append(
toil.importFile(
makeURL(options.blastOutput) +
'.ig_coverage_{}'.format(i)))
halID = toil.start(
Job.wrapJobFn(run_cactus_align, configWrapper, workFlowArgs,
project, cactus_blast_input))
# export the hal
toil.exportFile(halID, makeURL(options.outputHal))
def runCactusBlastOnly(options):
with Toil(options) as toil:
importSingularityImage(options)
#Run the workflow
if options.restart:
alignmentID = toil.restart()
else:
options.cactusDir = getTempDirectory()
# apply path overrides. this was necessary for wdl which doesn't take kindly to
# text files of local paths (ie seqfile). one way to fix would be to add support
# for s3 paths and force wdl to use it. a better way would be a more fundamental
# interface shift away from files of paths throughout all of cactus
if options.pathOverrides:
seqFile = SeqFile(options.seqFile)
configNode = ET.parse(options.configFile).getroot()
config = ConfigWrapper(configNode)
tree = MultiCactusTree(seqFile.tree)
tree.nameUnlabeledInternalNodes(
prefix=config.getDefaultInternalNodePrefix())
for name, override in zip(options.pathOverrideNames,
options.pathOverrides):
seqFile.pathMap[name] = override
override_seq = os.path.join(options.cactusDir,
'seqFile.override')
with open(override_seq, 'w') as out_sf:
out_sf.write(str(seqFile))
options.seqFile = override_seq
#to be consistent with all-in-one cactus, we make sure the project
#isn't limiting itself to the subtree (todo: parameterize so root can
#be passed through from prepare to blast/align)
proj_options = copy.deepcopy(options)
proj_options.root = None
#Create the progressive cactus project (as we do in runCactusProgressive)
projWrapper = ProjectWrapper(proj_options,
proj_options.configFile,
ignoreSeqPaths=options.root)
projWrapper.writeXml()
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)
# open up the experiment (as we do in ProgressiveUp.run)
# note that we copy the path into the options here
experimentFile = project.expMap[options.root]
expXml = ET.parse(experimentFile).getroot()
logger.info("Experiment {}".format(ET.tostring(expXml)))
experiment = ExperimentWrapper(expXml)
configPath = experiment.getConfigPath()
configXml = ET.parse(configPath).getroot()
seqIDMap = dict()
tree = MultiCactusTree(experiment.getTree()).extractSubTree(
options.root)
leaves = tree.getChildNames(tree.getRootName())
outgroups = experiment.getOutgroupGenomes()
genome_set = set(leaves + outgroups)
logger.info("Genomes in blastonly, {}: {}".format(
options.root, list(genome_set)))
print(str(project.inputSequenceMap))
#import the sequences (that we need to align for the given event, ie leaves and outgroups)
for genome, seq in list(project.inputSequenceMap.items()):
if genome in genome_set:
if os.path.isdir(seq):
tmpSeq = getTempFile()
catFiles([
os.path.join(seq, subSeq)
for subSeq in os.listdir(seq)
], tmpSeq)
seq = tmpSeq
seq = makeURL(seq)
project.inputSequenceIDMap[genome] = toil.importFile(seq)
else:
# out-of-scope sequences will only cause trouble later on
del project.inputSequenceMap[genome]
#import cactus config
if options.configFile:
cactusConfigID = toil.importFile(makeURL(options.configFile))
else:
cactusConfigID = toil.importFile(
makeURL(project.getConfigPath()))
project.setConfigID(cactusConfigID)
project.syncToFileStore(toil)
configNode = ET.parse(project.getConfigPath()).getroot()
configWrapper = ConfigWrapper(configNode)
configWrapper.substituteAllPredefinedConstantsWithLiterals()
workFlowArgs = CactusWorkflowArguments(
options,
experimentFile=experimentFile,
configNode=configNode,
seqIDMap=project.inputSequenceIDMap)
outWorkFlowArgs = toil.start(
CactusTrimmingBlastPhase(standAlone=True,
cactusWorkflowArguments=workFlowArgs,
phaseName="trimBlast"))
# export the alignments
toil.exportFile(outWorkFlowArgs.alignmentsID,
makeURL(options.outputFile))
# optional secondary alignments
if outWorkFlowArgs.secondaryAlignmentsID:
toil.exportFile(outWorkFlowArgs.secondaryAlignmentsID,
makeURL(options.outputFile) + '.secondary')
# outgroup fragments and coverage are necessary for cactus-align, as the sequence names got changed in the above alignemnts
for i, outgroupFragmentID in enumerate(
outWorkFlowArgs.outgroupFragmentIDs):
toil.exportFile(
outgroupFragmentID,
makeURL(options.outputFile) + '.og_fragment_{}'.format(i))
# cactus-align can recompute coverage on the fly, but we save them because we have them
for i, ingroupCoverageID in enumerate(
outWorkFlowArgs.ingroupCoverageIDs):
toil.exportFile(
ingroupCoverageID,
makeURL(options.outputFile) + '.ig_coverage_{}'.format(i))
def main():
args = initParser()
myProj = MultiCactusProject()
myProj.readXML(args['cactus_project'])
if not args['append']:
# Overwrite existing hal
print 'rm -f {0}'.format(args['HAL_file_path'])
system('rm -f {0}'.format(args['HAL_file_path']))
# some quick stats
totalTime = time.time()
totalAppendTime = 0
# traverse tree to make sure we are going breadth-first
tree = myProj.mcTree
# find subtree if event specified
event = args['event']
rootNode = None
if event is not None:
assert event in tree.nameToId and not tree.isLeaf(tree.nameToId[event])
rootNode = tree.nameToId[event]
for node in tree.breadthFirstTraversal(rootNode):
genomeName = tree.getName(node)
if genomeName in myProj.expMap:
experimentFilePath = myProj.expMap[genomeName]
print experimentFilePath
experiment = ExperimentWrapper(ET.parse(experimentFilePath).getroot())
outgroups = experiment.getOutgroupEvents()
expTreeString = NXNewick().writeString(experiment.getTree(onlyThisSubtree=True))
assert len(expTreeString) > 1
assert experiment.getHALPath() is not None
assert experiment.getHALFastaPath() is not None
cmdline = "time halAppendCactusSubtree \'{0}\' \'{1}\' \'{2}\' \'{3}\'".format(experiment.getHALPath(), experiment.getHALFastaPath(), expTreeString, args['HAL_file_path'])
if len(outgroups) > 0:
cmdline += " --outgroups {0}".format(",".join(outgroups))
if args["cacheBytes"] is not None:
cmdline += " --cacheBytes {0}".format(args["cacheBytes"])
if args["cacheMDC"] is not None:
cmdline += " --cacheMDC {0}".format(args["cacheMDC"])
if args["cacheRDC"] is not None:
cmdline += " --cacheRDC {0}".format(args["cacheRDC"])
if args["cacheW0"] is not None:
cmdline += " --cacheW0 {0}".format(args["cacheW0"])
if args["chunk"] is not None:
cmdline += " --chunk {0}".format(args["chunk"])
if args["deflate"] is not None:
cmdline += " --deflate {0}".format(args["deflate"])
if args["inMemory"] is True:
cmdline += " --inMemory"
print cmdline
appendTime = time.time()
system(cmdline)
appendTime = time.time() - appendTime
totalAppendTime += appendTime
# print "time of above command: {0:.2f}".format(appendTime)
totalTime = time.time() - totalTime
print "total time: {0:.2f} total halAppendCactusSubtree time: {1:.2f}".format(totalTime, totalAppendTime)
def runCactusBlastOnly(options):
with Toil(options) as toil:
importSingularityImage(options)
#Run the workflow
if options.restart:
alignmentID = toil.restart()
else:
options.cactusDir = getTempDirectory()
#Create the progressive cactus project (as we do in runCactusProgressive)
projWrapper = ProjectWrapper(options,
options.configFile,
ignoreSeqPaths=options.root)
projWrapper.writeXml()
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)
# open up the experiment (as we do in ProgressiveUp.run)
# note that we copy the path into the options here
experimentFile = project.expMap[options.root]
expXml = ET.parse(experimentFile).getroot()
logger.info("Experiment {}".format(ET.tostring(expXml)))
experiment = ExperimentWrapper(expXml)
configPath = experiment.getConfigPath()
configXml = ET.parse(configPath).getroot()
seqIDMap = dict()
tree = MultiCactusTree(experiment.getTree()).extractSubTree(
options.root)
leaves = tree.getChildNames(tree.getRootName())
outgroups = experiment.getOutgroupGenomes()
genome_set = set(leaves + outgroups)
logger.info("Genomes in blastonly, {}: {}".format(
options.root, list(genome_set)))
#import the sequences (that we need to align for the given event, ie leaves and outgroups)
for genome, seq in list(project.inputSequenceMap.items()):
if genome in genome_set:
if os.path.isdir(seq):
tmpSeq = getTempFile()
catFiles([
os.path.join(seq, subSeq)
for subSeq in os.listdir(seq)
], tmpSeq)
seq = tmpSeq
seq = makeURL(seq)
project.inputSequenceIDMap[genome] = toil.importFile(seq)
else:
# out-of-scope sequences will only cause trouble later on
del project.inputSequenceMap[genome]
#import cactus config
if options.configFile:
cactusConfigID = toil.importFile(makeURL(options.configFile))
else:
cactusConfigID = toil.importFile(
makeURL(project.getConfigPath()))
project.setConfigID(cactusConfigID)
project.syncToFileStore(toil)
configNode = ET.parse(project.getConfigPath()).getroot()
configWrapper = ConfigWrapper(configNode)
configWrapper.substituteAllPredefinedConstantsWithLiterals()
workFlowArgs = CactusWorkflowArguments(
options,
experimentFile=experimentFile,
configNode=configNode,
seqIDMap=project.inputSequenceIDMap)
outWorkFlowArgs = toil.start(
CactusTrimmingBlastPhase(standAlone=True,
cactusWorkflowArguments=workFlowArgs,
phaseName="trimBlast"))
# export the alignments
toil.exportFile(outWorkFlowArgs.alignmentsID,
makeURL(options.outputFile))
# optional secondary alignments
if outWorkFlowArgs.secondaryAlignmentsID:
toil.exportFile(outWorkFlowArgs.secondaryAlignmentsID,
makeURL(options.outputFile) + '.secondary')
# outgroup fragments and coverage are necessary for cactus-align, as the sequence names got changed in the above alignemnts
for i, outgroupFragmentID in enumerate(
outWorkFlowArgs.outgroupFragmentIDs):
toil.exportFile(
outgroupFragmentID,
makeURL(options.outputFile) + '.og_fragment_{}'.format(i))
# cactus-align can recompute coverage on the fly, but we save them because we have them
for i, ingroupCoverageID in enumerate(
outWorkFlowArgs.ingroupCoverageIDs):
toil.exportFile(
ingroupCoverageID,
makeURL(options.outputFile) + '.ig_coverage_{}'.format(i))
