def run(self):
logger.info('CycleStep4 object running, %s' % self.thisDir)
lsc.verifyDirExists(self.thisDir)
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'cycle.step4.start.xml'))
# lsc.subTypeTimestamp(self.thisDir, 'cycle', 'CycleStep4_start')
outname = os.path.join(self.thisDir, 'logs', 'gene_deactivation.log')
if not os.path.exists(outname):
if not self.options.noGeneDeactivation:
# by default gene deactivation is turned on.
cmd = lsc.evolverGeneDeactivationStep(self.thisDir, self.thisParentDir)
p = subprocess.Popen(cmd, cwd = self.getLocalTempDir(),
stdout = subprocess.PIPE, stderr = subprocess.STDOUT)
out = p.communicate()[0]
f=open(outname + '.tmp', 'w')
f.write(out)
f.close()
os.rename(outname + '.tmp', outname)
else:
# this could cause a proliferation of gene creation.
cmd = [lsc.which('cp')]
cmd.append(os.path.join(thisDir, 'intra', 'evannots.gff'))
cmd.append(os.path.join(thisDir, 'annots.gff'))
cmds = [cmd]
cmds.append([lsc.which('touch'), outname])
lsc.runCommands(cmds, self.getLocalTempDir())
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'cycle.step4.end.xml'))
def run(self):
if self.thisParentDir is None:
# happens when thisParentDir is the root
return
lsc.verifyDirExists(self.thisDir)
self.addChildTarget(StatsStep1(self.thisDir, self.thisParentDir, self.options))
self.setFollowOnTarget(Transalign(self.thisDir, self.thisParentDir, self.options))
def run(self):
chrNameDict, revChrNameDict = lsc.extractChrNamesDict(self.thisDir)
logger.info('CycleStep2Chromosome object running, %s %s %s'
% (self.thisDir, self.thisChr, chrNameDict[self.thisChr]))
lsc.verifyDirExists(self.thisDir)
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'cycle.step2.%s.start.xml'
% chrNameDict[self.thisChr]),
extra = {'name': self.thisChr})
# evolver intra on one chromosome
cmds = lsc.evolverIntraStepCmd(self.thisDir, self.theChild, self.thisStepLength,
self.thisChr, self.options.seed,
self.options.paramsDir, self.getLocalTempDir(), self.options)
lsc.runCommands(cmds, self.getLocalTempDir())
# evolver conversion from .rev to fasta in localTempDir
cmds = lsc.evolverIntraStepToFastaCmd(self.thisDir, self.thisStepLength, self.thisChr,
self.options.paramsDir, self.getLocalTempDir())
lsc.runCommands(cmds, self.getLocalTempDir())
# trf wrapper
lsc.callEvolverIntraStepTRFCmd(self.thisDir, self.thisChr, self.getLocalTempDir())
# move the resulting trf files out of localTempDir
cmds = lsc.evolverIntraStepMoveTRFCmd(self.thisDir, self.thisChr, self.getLocalTempDir())
lsc.runCommands(cmds, self.getLocalTempDir(), mode = 'p')
lsc.createTimestamp(os.path.join(self.thisDir, 'xml',
'cycle.step2.%s.end.xml' % chrNameDict[self.thisChr]),
extra = {'name': self.thisChr})
def run(self):
logger.info('Transalign object running, thisDir: %s thisParentDir: %s'
% (self.thisDir, self.thisParentDir))
if self.thisParentDir is None:
# happens when thisParentDir is the root
return
lsc.verifyDirExists(self.thisDir)
self.addChildTarget(TransalignStep(self.thisDir, self.thisParentDir, self.options))
def run(self):
logger.info('StatsStep3 object running, %s' % self.thisDir)
lsc.verifyDirExists(self.thisDir)
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'stats.step3.start.xml'))
cmds, pipes = lsc.statsStep3Cmds(self.thisDir, self.thisParentDir, self.options)
lsc.runCommands(cmds, self.getLocalTempDir(), outPipes = pipes)
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'stats.step3.end.xml'))
self.setFollowOnTarget(StatsStep4(self.thisDir, self.thisParentDir, self.options))
def run(self):
logger.info('StatsStep4 object running, %s' % self.thisDir)
lsc.verifyDirExists(self.thisDir)
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'stats.step4.start.xml'))
cmds, pipes = lsc.statsStep4Cmds(self.thisDir, self.thisParentDir, self.options)
lsc.runCommands(cmds, self.getLocalTempDir(), outPipes = pipes)
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'stats.step4.end.xml'))
if lsc.isLeaf(self.thisDir):
lsc.lastOneOutTurnOffTheLightsSimulation(self.options.simDir, self.options)
def run(self):
logger.info('CycleStep1 object running, %s' % self.thisDir)
lsc.verifyDirExists(self.thisDir)
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'cycle.step1.start.xml'))
lsc.runEvolverInterCmds(self.thisDir, self.thisParentDir, self.theChild, self.theParent,
self.thisStepLength, self.options.seed, self.options.paramsDir,
self.getLocalTempDir(), self.options)
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'cycle.step1.end.xml'))
self.setFollowOnTarget(CycleStep2(self.thisNewickStr, self.thisParentDir,
self.thisStepLength, self.options))
def run(self):
logger.info('CycleStep2 object running, %s' % self.thisDir)
lsc.verifyDirExists(self.thisDir)
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'cycle.step2.start.xml'))
lsc.verifyDirExists(os.path.join(self.thisDir, 'inter'))
lsc.verifyFileExists(os.path.join(self.thisDir, 'inter', 'inter.chrnames.txt'))
f = open(os.path.join(self.thisDir, 'inter', 'inter.chrnames.txt'), 'r')
for chrom in f:
chrom = chrom.strip()
self.addChildTarget(CycleStep2Chromosome(self.thisNewickStr, self.thisParentDir,
self.thisStepLength, chrom, self.options))
f.close()
self.setFollowOnTarget(CycleStep2FollowOn(self.thisNewickStr, self.thisParentDir,
self.thisStepLength, self.options))
def run(self):
logger.info('TransalignStep object running, thisDir: %s thisParentDir: %s'
% (self.thisDir, self.thisParentDir))
lsc.verifyDirExists(self.thisDir)
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'transalign.start.xml'))
cmds, pipes = lsc.transalignStepCmds_1(self.thisDir, self.thisParentDir, self.options)
lsc.runCommands(cmds, self.getLocalTempDir(), outPipes = pipes)
lsc.runTransalignStepCmds_2(self.thisDir, self.thisParentDir,
self.getLocalTempDir(), self.options)
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'transalign.end.xml'))
lsc.lastOneOutTurnOffTheLightsCycle(self.thisDir)
if lsc.isLeaf(self.thisDir):
lsc.lastOneOutTurnOffTheLightsSimulation(self.options.simDir, self.options)
def run(self):
logger.info('CycleStep3 object running, %s' % self.thisDir)
lsc.verifyDirExists(self.thisDir)
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'cycle.step3.start.xml'))
# trf
regex = r'^(chr\S+)\.outseq\.fa.*\.dat'
pat = re.compile(regex)
files = glob.glob(os.path.join(self.thisDir, 'intra', '*.dat'))
cmds = []
outPipes = []
followCmds = []
followPipes = []
for f in files:
# each file is the trf output for one chromosome
m = re.match(regex, os.path.basename(f))
if m is None:
raise RuntimeError('Regex "%s" failed on filename %s' % (regex, os.path.basename(f)))
outname = os.path.join(self.thisDir, 'intra', m.group(1) + '.trfannots.gff')
if not os.path.exists(outname):
# convert the .dat to .gff
cmd = [lsc.which('evolver_trf2gff.py'), f]
cmds.append(cmd)
outPipes.append(outname + '.tmp')
# atomic files
followCmds.append([lsc.which('mv'), outname + '.tmp', outname])
followPipes.append(None)
lsc.runCommands(cmds, self.getLocalTempDir(), outPipes = outPipes, mode = 'p')
lsc.runCommands(followCmds, self.getLocalTempDir(), outPipes = followPipes, mode = 'p')
catCmd, evoCmd, cvtCmd, followCmds = lsc.evolverIntraMergeCmds(self.thisDir, self.theChild)
lsc.runCommands([catCmd, evoCmd, cvtCmd], self.getLocalTempDir(),
outPipes = [os.path.join(self.thisDir, 'intra', 'evannots.gff.tmp'), None, None],
mode = 'p')
lsc.runCommands(followCmds, self.getLocalTempDir())
lsc.createTimestamp(os.path.join(self.thisDir, 'xml', 'cycle.step3.end.xml'))
self.setFollowOnTarget(CycleStep4(self.thisNewickStr, self.thisParentDir,
self.thisStepLength, self.options))
def checkForFiles(options):
""" If files are missing, complains and dies.
"""
lsc.verifyDirExists(os.path.join(options.rootInputDir, 'stats'))
for f in [os.path.join(options.rootInputDir, 'seq.rev'),
os.path.join(options.rootInputDir, 'annots.gff'),
os.path.join(options.paramsDir, 'model.txt'),
os.path.join(options.rootInputDir, 'stats', 'merged_root.stats.txt'),
os.path.join(options.rootInputDir, 'stats', 'cds_annots.gff'),
os.path.join(options.rootInputDir, 'stats', 'exons.gff'),
os.path.join(options.rootInputDir, 'stats', 'introns.gff'),
os.path.join(options.rootInputDir, 'stats', 'expanded_annots.gff'),
os.path.join(options.rootInputDir, 'stats', 'annotstats.txt')]:
lsc.verifyFileExists(f)
if not options.noMEs:
lsc.verifyDirExists(os.path.join(options.rootInputDir, 'mobiles'))
for f in [os.path.join(options.paramsDir, 'model.mes.txt'),
os.path.join(options.paramsDir, 'mes.cfg'),
os.path.join(options.rootInputDir, 'mobiles', 'LTR.fa'),
os.path.join(options.rootInputDir, 'mobiles', 'ME.fa'),
os.path.join(options.rootInputDir, 'mobiles', 'ME.gff')]:
lsc.verifyFileExists(f)