def runCactusBlastIngroupsAndOutgroups(ingroups, outgroups, alignmentsFile, toilDir, outgroupFragmentPaths=None, ingroupCoveragePaths=None, chunkSize=None, overlapSize=None,
logLevel=None,
compressFiles=None,
lastzMemory=None):
options = Job.Runner.getDefaultOptions(toilDir)
options.disableCaching = True
options.logLevel = "CRITICAL"
blastOptions = BlastOptions(chunkSize=chunkSize, overlapSize=overlapSize,
compressFiles=compressFiles,
memory=lastzMemory)
with Toil(options) as toil:
ingroupIDs = [toil.importFile(makeURL(ingroup)) for ingroup in ingroups]
outgroupIDs = [toil.importFile(makeURL(outgroup)) for outgroup in outgroups]
rootJob = BlastIngroupsAndOutgroups(blastOptions, ingroupIDs, outgroupIDs)
blastResults = toil.start(rootJob)
alignmentsID = blastResults[0]
toil.exportFile(alignmentsID, makeURL(alignmentsFile))
outgroupFragmentIDs = blastResults[1]
ingroupCoverageIDs = blastResults[2]
if outgroupFragmentPaths:
assert len(outgroupFragmentIDs) == len(outgroupFragmentPaths)
for outgroupFragmentID, outgroupFragmentPath in zip(outgroupFragmentIDs, outgroupFragmentPaths):
toil.exportFile(outgroupFragmentID, makeURL(outgroupFragmentPath))
if ingroupCoveragePaths:
assert len(ingroupCoverageIDs) == len(ingroupCoveragePaths)
for ingroupCoverageID, ingroupCoveragePath in zip(ingroupCoverageIDs, ingroupCoveragePaths):
toil.exportFile(ingroupCoverageID, makeURL(ingroupCoveragePath))
def runCactusBlast(sequenceFiles,
alignmentsFile,
toilDir,
chunkSize=None,
overlapSize=None,
logLevel=None,
compressFiles=None,
lastzMemory=None,
targetSequenceFiles=None):
options = Job.Runner.getDefaultOptions(toilDir)
options.logLevel = "CRITICAL"
blastOptions = BlastOptions(chunkSize=chunkSize,
overlapSize=overlapSize,
compressFiles=compressFiles,
memory=lastzMemory)
with Toil(options) as toil:
seqIDs = [
toil.importFile(makeURL(seqFile)) for seqFile in sequenceFiles
]
if targetSequenceFiles:
targetSeqIDs = [
toil.importFile(makeURL(seqFile))
for seqFile in targetSequenceFiles
]
rootJob = BlastSequencesAgainstEachOther(
sequenceFileIDs1=seqIDs,
sequenceFileIDs2=targetSeqIDs,
blastOptions=blastOptions)
else:
rootJob = BlastSequencesAllAgainstAll(seqIDs, blastOptions)
alignmentsID = toil.start(rootJob)
toil.exportFile(alignmentsID, makeURL(alignmentsFile))