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():
name = tree.getName(node)
if tree.isLeaf(node) or (name == experiment.getRootGenome() and experiment.isRootReconstructed() == False):
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
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.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
class TestCase(unittest.TestCase):
def setUp(self):
unittest.TestCase.setUp(self)
self.tree = NXNewick().parseString(
'((((HUMAN:0.006969,CHIMP:0.009727)anc2:0.025291,BABOON:0.044568)anc1:0.11,(MOUSE:0.072818,RAT:0.081244):0.260342):0.02326,((DOG:0.07,CAT:0.07):0.087381,(PIG:0.06,COW:0.06):0.104728):0.04);'
)
self.xmlRoot = self.__makeXmlDummy()
self.exp = ExperimentWrapper(self.xmlRoot)
self.exp.setTree(self.tree)
self.seqMap = {
'HUMAN': 'human.txt',
'CHIMP': 'chimp.txt',
'BABOON': 'baboon.txt',
'MOUSE': 'mouse.txt',
'RAT': 'rat.txt',
'DOG': 'dog.txt',
'CAT': 'cat.txt',
'PIG': 'pig.txt',
'COW': 'cow.txt'
}
self.exp.setRootGenome('anc1')
self.exp.setRootReconstructed(True)
self.exp.setOutgroupGenomes(
['MOUSE', 'RAT', 'DOG', 'CAT', 'PIG', 'COW'])
for genome, seq in self.seqMap.items():
# These aren't real IDs, but should still work for our
# purposes
self.exp.setSequenceID(genome, seq)
def testGetSequencePath(self):
for genome, seq in self.seqMap.items():
self.assertEqual(self.exp.getSequenceID(genome), seq)
# Should not be any entries for genomes not in the tree
self.assertEqual(self.exp.getSequenceID('DUCK'), None)
def testChangingSequencePaths(self):
"""Tests that changing a sequence path persists correctly."""
self.exp.setSequenceID('HUMAN', 'human2.txt')
self.assertEqual(self.exp.getSequenceID('HUMAN'), 'human2.txt')
# Reload the wrapper and try again
self.exp = ExperimentWrapper(self.xmlRoot)
self.assertEqual(self.exp.getSequenceID('HUMAN'), 'human2.txt')
def testOutgroups(self):
self.assertEqual(set(self.exp.getOutgroupGenomes()),
set(['MOUSE', 'RAT', 'DOG', 'CAT', 'PIG', 'COW']))
self.exp.setOutgroupGenomes([])
self.assertEqual(self.exp.getOutgroupGenomes(), [])
def testRootGenome(self):
self.assertEqual(self.exp.getRootGenome(), 'anc1')
self.exp.setRootGenome('anc2')
self.assertEqual(self.exp.getRootGenome(), 'anc2')
def testSetTree(self):
# A modfied version, with fewer genomes and a new one
tree2 = NXNewick().parseString(
'((HUMAN:0.006969,CHIMP:0.009727):0.025291,BABOON:0.044568,ARMADILLO:1.0);'
)
self.exp.setTree(tree2)
self.assertEqual(set(self.exp.getGenomesWithSequence()),
set(['HUMAN', 'CHIMP', 'BABOON']))
def __makeXmlDummy(self):
rootElem = ET.Element("dummy")
rootElem.append(self.__makeDiskElem())
return rootElem
def __makeDiskElem(self):
diskElem = ET.Element("cactus_disk")
confElem = ET.Element("st_kv_database_conf")
confElem.attrib['type'] = 'kyoto_tycoon'
diskElem.append(confElem)
dbElem = ET.Element('kyoto_tycoon')
confElem.append(dbElem)
return diskElem