def makeMaf(inHalPath, outDir, step, overwrite, doMaf):
srcHalPath = inHalPath
if step > 0:
srcHalPath = makePath(inHalPath, outDir, step, "lod", "hal")
outMafPath = makePath(inHalPath, outDir, step, "out", "maf")
if doMaf and (overwrite or not os.path.isfile(outMafPath)):
runShellCommand("hal2maf %s %s" % (srcHalPath, outMafPath))
def runSteps(inHalPath, outDir, maxBlock, scale, steps, overwrite, doMaf,
keepSeq, trans, inMemory):
table = defaultdict(list)
makeMaf(inHalPath, outDir, 0, overwrite, doMaf)
table[0] = [os.path.getsize(inHalPath) / 1024]
table[0] += list(getHalTotalSegments(inHalPath))
table[0] += getPrecisionRecall(inHalPath, outDir, 0, False)
table[0] += getScanTime(inHalPath, outDir, 0)
if steps is None:
steps = getSteps(inHalPath, maxBlock, scale)
for stepIdx in xrange(1,len(steps)):
step = steps[stepIdx]
outPath = makePath(inHalPath, outDir, step, "lod", "hal")
srcPath = inHalPath
if trans is True and stepIdx > 1:
srcPath = makePath(inHalPath, outDir, steps[stepIdx-1],
"lod", "hal")
if overwrite is True or not os.path.isfile(outPath):
stepScale = (scale ** stepIdx)
runHalLodExtract(srcPath, outPath, stepScale, keepSeq, inMemory)
makeMaf(inHalPath, outDir, step, overwrite, doMaf)
compMaf(inHalPath, outDir, step, overwrite, doMaf)
table[step] = [os.path.getsize(outPath) / 1024]
table[step] += list(getHalTotalSegments(outPath))
table[step] += getPrecisionRecall(inHalPath, outDir, step, doMaf)
table[step] += getScanTime(inHalPath, outDir, step)
return table
def getPrecisionRecall(inHalPath, outDir, step, doMaf):
if doMaf:
sumPath = makePath(inHalPath, outDir, step, "comp", "txt")
sumFile = open(sumPath, "r")
line = sumFile.next()
line = sumFile.next()
line = sumFile.next()
tokens = line.split()
assert tokens[2] == "self)"
sumNearPath = makePath(inHalPath, outDir, step, "comp_near", "txt")
sumNearFile = open(sumNearPath, "r")
line = sumNearFile.next()
line = sumNearFile.next()
line = sumNearFile.next()
tokensNear = line.split()
assert tokensNear[2] == "self)"
return [
float(tokens[3]),
float(tokens[4]),
float(tokensNear[3]),
float(tokensNear[4])
]
elif step == 0:
return [1., 1., 1., 1.]
else:
return [0., 0., 0., 0.]
def getScanTime(inHalPath, outDir, step):
srcHalPath = inHalPath
if step > 0:
srcHalPath = makePath(inHalPath, outDir, step, "lod", "hal")
genomes = getHalGenomes(inHalPath)
assert len(genomes) > 1
genName = genomes[1]
bedPath = makePath(inHalPath, outDir, step, genName, "bed")
t1 = time.time()
runShellCommand("halBranchMutations %s %s --refFile %s" % (srcHalPath, genName, bedPath))
elapsedTime = time.time() - t1
return [elapsedTime]
def getScanTime(inHalPath, outDir, step):
srcHalPath = inHalPath
if step > 0:
srcHalPath = makePath(inHalPath, outDir, step, "lod", "hal")
genomes = getHalGenomes(inHalPath)
assert len(genomes) > 1
genName = genomes[1]
bedPath = makePath(inHalPath, outDir, step, genName, "bed")
t1 = time.time()
runShellCommand("halBranchMutations %s %s --refFile %s" % (
srcHalPath, genName, bedPath))
elapsedTime = time.time() - t1
return [elapsedTime]
def getPrecisionRecall(inHalPath, outDir, step, doMaf):
if doMaf:
sumPath = makePath(inHalPath, outDir, step, "comp", "txt")
sumFile = open(sumPath, "r")
line = sumFile.next()
line = sumFile.next()
line = sumFile.next()
tokens = line.split()
assert tokens[2] == "self)"
sumNearPath = makePath(inHalPath, outDir, step, "comp_near", "txt")
sumNearFile = open(sumNearPath, "r")
line = sumNearFile.next()
line = sumNearFile.next()
line = sumNearFile.next()
tokensNear = line.split()
assert tokensNear[2] == "self)"
return [float(tokens[3]), float(tokens[4]), float(tokensNear[3]), float(tokensNear[4])]
elif step == 0:
return [1.0, 1.0, 1.0, 1.0]
else:
return [0.0, 0.0, 0.0, 0.0]
def compMaf(inHalPath, outDir, step, overwrite, doMaf):
srcMaf = makePath(inHalPath, outDir, 0, "out", "maf")
tgtMaf = makePath(inHalPath, outDir, step, "out", "maf")
xmlPath = makePath(inHalPath, outDir, step, "comp", "xml")
sumPath = makePath(inHalPath, outDir, step, "comp", "txt")
if doMaf and (overwrite or not os.path.isfile(xmlPath)):
runShellCommand("mafComparator --maf1 %s --maf2 %s --out %s --samples 100000" % (srcMaf, tgtMaf, xmlPath))
runShellCommand("comparatorSummarizer.py --xml %s > %s " % (xmlPath, sumPath))
xmlNearPath = makePath(inHalPath, outDir, step, "comp_near", "xml")
sumNearPath = makePath(inHalPath, outDir, step, "comp_near", "txt")
if doMaf and (overwrite or not os.path.isfile(xmlNearPath)):
runShellCommand(
"mafComparator --maf1 %s --maf2 %s --out %s --near %d --samples 100000"
% (srcMaf, tgtMaf, xmlNearPath, int(step))
)
runShellCommand("comparatorSummarizer.py --xml %s > %s " % (xmlNearPath, sumNearPath))
def compMaf(inHalPath, outDir, step, overwrite, doMaf):
srcMaf = makePath(inHalPath, outDir, 0, "out", "maf")
tgtMaf = makePath(inHalPath, outDir, step, "out", "maf")
xmlPath = makePath(inHalPath, outDir, step, "comp", "xml")
sumPath = makePath(inHalPath, outDir, step, "comp", "txt")
if doMaf and (overwrite or not os.path.isfile(xmlPath)):
runShellCommand("mafComparator --maf1 %s --maf2 %s --out %s --samples 100000" % (
srcMaf, tgtMaf, xmlPath))
runShellCommand("comparatorSummarizer.py --xml %s > %s " % (xmlPath,
sumPath))
xmlNearPath = makePath(inHalPath, outDir, step, "comp_near", "xml")
sumNearPath = makePath(inHalPath, outDir, step, "comp_near", "txt")
if doMaf and (overwrite or not os.path.isfile(xmlNearPath)):
runShellCommand(
"mafComparator --maf1 %s --maf2 %s --out %s --near %d --samples 100000" % (
srcMaf, tgtMaf, xmlNearPath, int(step)))
runShellCommand("comparatorSummarizer.py --xml %s > %s " % (
xmlNearPath, sumNearPath))