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 readTurnoverFile(halPath, turnoverPath):
result = dict()
toFile = open(turnoverPath, "r")
for line in toFile:
toks = line.split()
genome = toks[0].strip(":")
cons = float(toks[2])
ucons = float(toks[4])
gain = float(toks[6])
loss = float(toks[9])
totalAligned = cons + ucons + gain + loss
if totalAligned <= 0 or cons < 0 or ucons < 0:
sys.stderr.write("Warning, skipping %s\n" % genome)
else:
pi0 = (ucons + loss) / totalAligned
pi1 = (cons + gain) / totalAligned
pg = gain / (ucons + gain)
pl = loss / (cons + loss)
t = float(toks[12])
#
# Incorporate parent branch since it affects turnover
#
if genome != getHalRootName(halPath):
parName = getParentGenomeName(halPath, genome)
parBranch = getBranchLength(halPath, parName)
t += float(parBranch)
assert pi0 >= 0 and pi1 >=0
assert pg >= 0 and pl >=0
assert t >= 0
result[genome] = ([pi0, pi1], [ [1.0 - pg, pg], [pl, 1.0 - pl] ], t)
return result
def main(argv=None):
if argv is None:
argv = sys.argv
parser = argparse.ArgumentParser()
parser.add_argument("halFile", type=str,
help="Path of hal file")
parser.add_argument("NITurnoverFile", type=str,
help="Output of halTreeNITurnover.py")
parser.add_argument("--maxIt", type=int, default=100000,
help="number of iterations for gradient descent")
parser.add_argument("--step", type=float, default=0.0001,
help="gradient descent step")
parser.add_argument("--retries", type=int, default=5,
help="number of gradient descents to run")
parser.add_argument("--root", type=str, default=None,
help="root of alignment to consder")
parser.add_argument("--allInternals", action="store_true", default=False,
help="estimate params for all subtrees independently,"
" in addition to the root")
args = parser.parse_args()
if args.root is None:
args.root = getHalRootName(args.halFile)
assert (args.maxIt > 0 and args.step > 0 and args.retries > 1)
halTreeTurnoverParams(args.halFile, args.NITurnoverFile,
args.root, args.allInternals, args.maxIt,
args.step, args.retries)
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)),
)