def halTreeTurnoverParams(halPath, obsPath, rootName, allInternals,
maxIt, step, retries):
observations = readTurnoverFile(halPath, obsPath)
nextQueue = deque()
nextQueue.append(rootName)
output = []
while len(nextQueue) > 0:
next = nextQueue.popleft()
if next == rootName or (allInternals is True and
len(getHalChildrenNames(halPath, next)) > 0):
obsVals = getValuesBelowRoot(halPath, next, observations)
result = estimateParamsFromList(obsVals, maxIt, step, retries)
print "%s: lr=%f gr=%f dsq=%f" % (next,result[0], result[1],
result[2])
printComparison(halPath, obsVals, observations, result)
for child in getHalChildrenNames(halPath, next):
nextQueue.append(child)
return output
def getHalTreeTurnover(halPath, args, rootName=None):
root = rootName
if root is None:
root = getHalRootName(halPath)
for child in getHalChildrenNames(halPath, root):
if root != getHalRootName(halPath):
consFile = os.path.join(args.workDir,
args.conservedBedName % child)
checkFile(consFile)
pconsFile = os.path.join(args.workDir,
args.conservedBedName % root)
checkFile(pconsFile)
outMappedAlignedBed = os.path.join(args.workDir,
child + "_pa.bed")
outParentSlicedBed = os.path.join(args.workDir,
child + "_pslice.bed")
outMappedGenomeBed = os.path.join(args.workDir,
child + "_pm.bed")
outConservationBed = os.path.join(args.workDir,
child + "_int.bed")
outAlignedBed = os.path.join(args.workDir, child + "_al.bed")
outGainBed = os.path.join(args.workDir, child + "_gain.bed")
outLossBed = os.path.join(args.workDir, child + "_loss.bed")
(conLen, gainLen,
lossLen, unconLen) = compareConservationOverBranch(
halPath, child, consFile, pconsFile,
outMappedAlignedBed, outParentSlicedBed,
outMappedGenomeBed, outConservationBed, outAlignedBed,
outGainBed, outLossBed)
gainRate = 0
if conLen + lossLen > 0:
gainRate = float(gainLen) / (unconLen + gainLen)
lossRate = 0
if unconLen + gainLen > 0:
lossRate = float(lossLen) / (conLen + lossLen)
branchLength = getBranchLength(halPath, child)
print "%s: cons %d ucons %d gain %d (%f) loss %d (%f) bl %f" % (
child,
conLen,
unconLen,
gainLen,
gainRate,
lossLen,
lossRate,
branchLength)
getHalTreeTurnover(halPath, args, child)
def getHalTreeBackground(halPath, args, rootName=None):
root = rootName
if root is None:
root = getHalRootName(halPath)
for child in getHalChildrenNames(halPath, root):
bgFile = os.path.join(args.workDir, args.backgroundBedName % child)
if args.ar is True:
command = "halMaskExtract %s %s --maskFile %s --extend %d --extendPct %f" % (halPath, child, bgFile, args.arExtend, args.arExtendPct)
else:
command = "halStats %s --bedSequences %s > %s" % (halPath, child,
bgFile)
print(command)
runShellCommand(command)
getHalTreeBackground(halPath, args, child)
def getValuesBelowRoot(halPath, rootName, observations):
nextQueue = deque()
nextQueue.append(rootName)
output = []
while len(nextQueue) > 0:
next = nextQueue.popleft()
if next != rootName:
if next in observations:
output.append(observations[next])
else:
sys.stderr.write("Warning, no observation for %s\n" % next)
for child in getHalChildrenNames(halPath, next):
nextQueue.append(child)
return output
def getHalTreeBackground(halPath, args, rootName=None):
root = rootName
if root is None:
root = getHalRootName(halPath)
for child in getHalChildrenNames(halPath, root):
bgFile = os.path.join(args.workDir, args.backgroundBedName % child)
if args.ar is True:
command = "halMaskExtract %s %s --maskFile %s --extend %d --extendPct %f" % (halPath, child, bgFile, args.arExtend, args.arExtendPct)
else:
command = "halStats %s --bedSequences %s > %s" % (halPath, child,
bgFile)
print command
runShellCommand(command)
getHalTreeBackground(halPath, args, child)
def getHalTreeConservation(halPath, args, events, rootName=None):
root = rootName
if root is None:
root = getHalRootName(halPath)
for child in getHalChildrenNames(halPath, root):
bgFile = os.path.join(args.workDir, args.backgroundBedName % child)
muFile = os.path.join(args.workDir, args.mutationsBedName % child)
checkFiles(bgFile, muFile)
outPath = os.path.join(args.workDir, args.conservedBedName % child)
outFile = open(outPath, "w")
bc = BedConservation()
bc.computeBackgroundRate(muFile, bgFile, events)
bc.identifyConservedIntervals(muFile, outFile, float(args.pval),
float(args.cutoff))
getHalTreeConservation(halPath, args, events, child)
print "%s: %d segments with %d bases (%f pct of genome) found. bgrate= %f minDist=%d" % (
child, bc.writtenCount, bc.writtenBases,
float(bc.writtenBases) / float(genomeLength(halPath, child)),
bc.rate, bc.minDistance(float(args.pval)))
def getHalTreeTurnover(halPath, args, rootName=None):
root = rootName
if root is None:
root = getHalRootName(halPath)
for child in getHalChildrenNames(halPath, root):
if root != getHalRootName(halPath):
consFile = os.path.join(args.workDir,
args.conservedBedName % child)
checkFile(consFile)
pconsFile = os.path.join(args.workDir,
args.conservedBedName % root)
checkFile(pconsFile)
outMappedAlignedBed = os.path.join(args.workDir, child + "_pa.bed")
outParentSlicedBed = os.path.join(args.workDir,
child + "_pslice.bed")
outMappedGenomeBed = os.path.join(args.workDir, child + "_pm.bed")
outConservationBed = os.path.join(args.workDir, child + "_int.bed")
outAlignedBed = os.path.join(args.workDir, child + "_al.bed")
outGainBed = os.path.join(args.workDir, child + "_gain.bed")
outLossBed = os.path.join(args.workDir, child + "_loss.bed")
(conLen, gainLen,
lossLen, unconLen) = compareConservationOverBranch(
halPath, child, consFile, pconsFile, outMappedAlignedBed,
outParentSlicedBed, outMappedGenomeBed, outConservationBed,
outAlignedBed, outGainBed, outLossBed)
gainRate = 0
if conLen + lossLen > 0:
gainRate = float(gainLen) / (unconLen + gainLen)
lossRate = 0
if unconLen + gainLen > 0:
lossRate = float(lossLen) / (conLen + lossLen)
branchLength = getBranchLength(halPath, child)
print "%s: cons %d ucons %d gain %d (%f) loss %d (%f) bl %f" % (
child, conLen, unconLen, gainLen, gainRate, lossLen, lossRate,
branchLength)
getHalTreeTurnover(halPath, args, child)
def getHalTreeConservation(halPath, args, events, rootName=None):
root = rootName
if root is None:
root = getHalRootName(halPath)
for child in getHalChildrenNames(halPath, root):
bgFile = os.path.join(args.workDir, args.backgroundBedName % child)
muFile = os.path.join(args.workDir, args.mutationsBedName % child)
checkFiles(bgFile, muFile)
outPath = os.path.join(args.workDir, args.conservedBedName % child)
outFile = open(outPath, "w")
bc = BedConservation()
bc.computeBackgroundRate(muFile, bgFile, events)
bc.identifyConservedIntervals(muFile, outFile, float(args.pval), float(args.cutoff))
getHalTreeConservation(halPath, args, events, child)
print "%s: %d segments with %d bases (%f pct of genome) found. bgrate= %f minDist=%d" % (
child,
bc.writtenCount,
bc.writtenBases,
float(bc.writtenBases) / float(genomeLength(halPath, child)),
bc.rate,
bc.minDistance(float(args.pval)),
)