def plotASProfile(tcc, cName, directory = None, min = 0, extra = "0"):
if not directory:
fN = extra + '.' + tcc + '.png'
else:
fN = directory + '/' + extra + '.' + tcc + '.png'
#Get S Profile
tccStretch = cgPeaks.stretch(tcc, cName)
highest = tccStretch.getHighestLevel()
if highest < min:
return 0
sortedX = tccStretch.profile.keys()
sortedX.sort()
sortedY = []
for X in sortedX:
sortedY.append(tccStretch.profile[X])
#Get AS Profile
chr, strand, start, end = tcc.strip().split(':')
if strand == '1':
strand = '-1'
else:
strand = '1'
tcc = cg.makeTcc(chr, strand, start, end)
tccStretchAS = cgPeaks.stretch(tcc, cName)
highest = tccStretchAS.getHighestLevel()
if highest < min:
return 0 #AS can have minimum I guess...
sortedXAS = tccStretchAS.profile.keys()
sortedXAS.sort()
sortedYAS = []
for X in sortedXAS:
sortedYAS.append(tccStretchAS.profile[X])
#Plot them
gDevice = importr('grDevices')
gDevice.png(file=fN, width=1680, height=1050)
r('split.screen(c(2,1))')
r('screen(1)')
r.plot(sortedX, sortedY, xlab = "Coordinates", ylab = "(Syn) Expression Level" )
r.lines(sortedX, sortedY, type = "b")
r('screen(2)')
r.plot(sortedXAS, sortedYAS, xlab = "Coordinates", ylab = "(Anti) Expression Level")
r.lines(sortedXAS, sortedYAS, type = "b")
gDevice.dev_off()
def profileTargetsHistoAS(tccList, cName, name='boxplot'):
range = 50
histDict = {} # {coord: []}
histDictAS = {}
for tcc in tccList:
chrom, strand, start, end = cg.tccSplit(tcc)
#Get highest peak (sense)
tccStretch = cgPeaks.stretch(tcc, cName)
tccStretch.createPeaks(span=2)
highestCoord = tccStretch.getHighestPeak()
if highestCoord == None: continue
#AS
tccAS = cg.convertToAS(tcc)
tccStretch = cgPeaks.stretch(tccAS, cName)
tccStretch.createPeaks(span=2)
highestCoordAS = tccStretch.getHighestPeak()
if highestCoordAS == None: continue
#profile around point (Sense)
zPoint = cg.makeTcc(chrom, strand, highestCoord, end)
cProfile = svs.profileAroundPoint(zPoint, range, cName, ratio=True)
for coord in cProfile:
try:
histDict[coord].append(cProfile[coord])
except: #quicker way to initialize
histDict[coord] = [cProfile[coord]]
#profile around point (AS)
zPoint = cg.convertToAS(zPoint)
cProfile = svs.profileAroundPoint(zPoint,
range,
cName,
ratio=True,
ratioCoord=highestCoordAS)
for coord in cProfile:
try:
histDictAS[coord].append(cProfile[coord])
except: #quicker way to initialize
histDictAS[coord] = [cProfile[coord]]
plot.boxPlotHistoAS(histDict, histDictAS, name=name)
def profileTargetsHistoAS(tccList, cName, name = 'boxplot'):
range = 50
histDict = {} # {coord: []}
histDictAS = {}
for tcc in tccList:
chrom, strand, start, end = cg.tccSplit(tcc)
#Get highest peak (sense)
tccStretch = cgPeaks.stretch(tcc, cName)
tccStretch.createPeaks(span = 2)
highestCoord = tccStretch.getHighestPeak()
if highestCoord == None: continue
#AS
tccAS = cg.convertToAS(tcc)
tccStretch = cgPeaks.stretch(tccAS, cName)
tccStretch.createPeaks(span = 2)
highestCoordAS = tccStretch.getHighestPeak()
if highestCoordAS == None: continue
#profile around point (Sense)
zPoint = cg.makeTcc(chrom, strand, highestCoord, end)
cProfile = svs.profileAroundPoint(zPoint, range, cName, ratio = True)
for coord in cProfile:
try:
histDict[coord].append(cProfile[coord])
except: #quicker way to initialize
histDict[coord] = [cProfile[coord]]
#profile around point (AS)
zPoint = cg.convertToAS(zPoint)
cProfile = svs.profileAroundPoint(zPoint, range, cName, ratio = True, ratioCoord = highestCoordAS)
for coord in cProfile:
try:
histDictAS[coord].append(cProfile[coord])
except: #quicker way to initialize
histDictAS[coord] = [cProfile[coord]]
plot.boxPlotHistoAS(histDict, histDictAS, name = name)
def updateSmallExpression(oFN, cName, rn = None, tn = None):
oNX = cgNexusFlat.Nexus(oFN, cgOriginRNAFlat.OriginRNA)
oNX.load(['eLevel', 'tcc'], [rn, tn])
for oID in oNX.eLevel:
stretch = cgPeaks.stretch(oNX.tcc[oID], cName) #this stretch contains values for small library...
highValue = stretch.getHighestLevel()
oNX.eLevel[oID] = highValue
oNX.save()
def updateSmallExpression(aDir, cName):
oRNA_DC = cgDB.dataController(aDir, cgOriginRNA.OriginRNA)
id_oRNA = oRNA_DC.load()
for id, oRNA in id_oRNA.items():
stretch = cgPeaks.stretch(oRNA.tcc, cName) #this stretch contains values for small library...
highValue = stretch.getHighestLevel()
oRNA.eLevel = highValue
oRNA_DC.commit(id_oRNA)
def plotPairs(oDir, aDir, cName):
oDC = cgDB.dataController(oDir, cgOriginRNA.OriginRNA)
id_oRNA = oDC.load()
aDC = cgDB.dataController(aDir, cgAlignment.cgAlignment)
id_alignment = aDC.load()
for oID, oRNA in id_oRNA.items():
if not oRNA.passedFilter:
continue
for aID in oRNA.filteredTargets:
alignment = id_alignment[aID]
chrom, strand, start, end = bioLibCG.tccSplit(alignment.tTcc)
offset = alignment.tStart
sLen = alignment.sLength
print sLen
print oRNA.sequence
print oRNA.tcc
print alignment.tTcc
if strand == '1':
start = start - 19 + offset
end = start + sLen
else:
end = end + 19 - offset
start = end - sLen
print chrom, strand, start, end
scanRange = bioLibCG.makeTcc(chrom, strand, start, end)
stretch = cgPeaks.stretch(scanRange, cName)
sortedKeys = stretch.profile.keys()
sortedKeys.sort()
if strand == '-1':
sortedKeys.reverse()
xVals = range(1, sLen + 2)
xVals = sortedKeys
yVals = [stretch.profile[x] for x in sortedKeys]
print xVals, len(xVals)
print yVals, len(yVals)
plt.plot(xVals, yVals)
plt.show()
return 0
def plotSmallDeg(tcc, smallCName, degCName, outDir = None, description = "None", nameNum = "0"):
if not outDir:
fN = nameNum + "." + tcc + '.png'
else:
fN = outDir + '/' + nameNum + "." + tcc + '.png'
#Get deg Profile
tccStretch = cgPeaks.stretch(tcc, degCName)
sortedX = tccStretch.profile.keys()
sortedX.sort()
sortedY = []
for X in sortedX:
sortedY.append(tccStretch.profile[X])
#Get small
tccStretchSmall = cgPeaks.stretch(tcc, smallCName)
sortedXAS = tccStretchSmall.profile.keys()
sortedXAS.sort()
sortedYAS = []
for X in sortedXAS:
sortedYAS.append(tccStretchSmall.profile[X])
#Plot them
gDevice = importr('grDevices')
gDevice.png(file=fN, width=1680, height=1050)
r('split.screen(c(2,1))')
r('screen(1)')
r.plot(sortedX, sortedY, xlab = "Coordinates", ylab = "Degradome Expression" )
r.lines(sortedX, sortedY, type = "b")
r('screen(2)')
r.plot(sortedXAS, sortedYAS, xlab = description, ylab = "Small Expression")
r.lines(sortedXAS, sortedYAS, type = "b")
gDevice.dev_off()
def plotPairs(oDir, aDir, cName):
oDC = cgDB.dataController(oDir, cgOriginRNA.OriginRNA)
id_oRNA = oDC.load()
aDC = cgDB.dataController(aDir, cgAlignment.cgAlignment)
id_alignment = aDC.load()
for oID, oRNA in id_oRNA.items():
if not oRNA.passedFilter:
continue
for aID in oRNA.filteredTargets:
alignment = id_alignment[aID]
chrom, strand, start, end = bioLibCG.tccSplit(alignment.tTcc)
offset = alignment.tStart
sLen = alignment.sLength
print sLen
print oRNA.sequence
print oRNA.tcc
print alignment.tTcc
if strand == '1':
start = start - 19 + offset
end = start + sLen
else:
end = end + 19 - offset
start = end - sLen
print chrom, strand, start, end
scanRange = bioLibCG.makeTcc(chrom, strand, start, end)
stretch = cgPeaks.stretch(scanRange, cName)
sortedKeys = stretch.profile.keys()
sortedKeys.sort()
if strand == '-1':
sortedKeys.reverse()
xVals = range(1, sLen + 2)
xVals = sortedKeys
yVals = [stretch.profile[x] for x in sortedKeys]
print xVals, len(xVals)
print yVals, len(yVals)
plt.plot(xVals, yVals)
plt.show()
return 0
def updateSmallExpression(oFN, cName, rn=None, tn=None):
oNX = cgNexusFlat.Nexus(oFN, cgOriginRNAFlat.OriginRNA)
oNX.load(['eLevel', 'tcc'], [rn, tn])
for oID in oNX.eLevel:
stretch = cgPeaks.stretch(
oNX.tcc[oID],
cName) #this stretch contains values for small library...
highValue = stretch.getHighestLevel()
oNX.eLevel[oID] = highValue
oNX.save()
def updateSmallExpression(aDir, cName):
oRNA_DC = cgDB.dataController(aDir, cgOriginRNA.OriginRNA)
id_oRNA = oRNA_DC.load()
for id, oRNA in id_oRNA.items():
stretch = cgPeaks.stretch(
oRNA.tcc,
cName) #this stretch contains values for small library...
highValue = stretch.getHighestLevel()
oRNA.eLevel = highValue
oRNA_DC.commit(id_oRNA)
def markCenterExpression(aDir, cName):
aDC = cgDB.dataController(aDir, cgAlignment.cgAlignment)
id_alignment = aDC.load()
for alignment in id_alignment.values():
alignment.centerExpression = [0.0, 0.0, 0.0]
chrom, strand, start, end = bioLibCG.tccSplit(alignment.tTcc)
offset = alignment.tStart
sLen = alignment.sLength
if strand == '1':
start = start - 19 + offset
end = start + sLen
else:
end = end + 19 - offset
start = end - sLen
scanRange = bioLibCG.makeTcc(chrom, strand, start, end)
stretch = cgPeaks.stretch(scanRange, cName)
expressionSum = stretch.getSumOfLevels()
sortedKeys = stretch.profile.keys()
sortedKeys.sort()
if strand == '-1':
sortedKeys.reverse()
if expressionSum != 0:
sum = 0.0
for key in sortedKeys[8:12]:
sum += stretch.profile[key]
alignment.centerExpression[0] = sum/expressionSum
sum = 0.0
for key in sortedKeys[7:13]:
sum += stretch.profile[key]
alignment.centerExpression[1] = sum/expressionSum
sum = 0.0
for key in sortedKeys[6:14]:
sum += stretch.profile[key]
alignment.centerExpression[2] = sum/expressionSum
aDC.commit(id_alignment)
def markCenterExpression(aDir, cName):
aDC = cgDB.dataController(aDir, cgAlignment.cgAlignment)
id_alignment = aDC.load()
for alignment in id_alignment.values():
alignment.centerExpression = [0.0, 0.0, 0.0]
chrom, strand, start, end = bioLibCG.tccSplit(alignment.tTcc)
offset = alignment.tStart
sLen = alignment.sLength
if strand == '1':
start = start - 19 + offset
end = start + sLen
else:
end = end + 19 - offset
start = end - sLen
scanRange = bioLibCG.makeTcc(chrom, strand, start, end)
stretch = cgPeaks.stretch(scanRange, cName)
expressionSum = stretch.getSumOfLevels()
sortedKeys = stretch.profile.keys()
sortedKeys.sort()
if strand == '-1':
sortedKeys.reverse()
if expressionSum != 0:
sum = 0.0
for key in sortedKeys[8:12]:
sum += stretch.profile[key]
alignment.centerExpression[0] = sum / expressionSum
sum = 0.0
for key in sortedKeys[7:13]:
sum += stretch.profile[key]
alignment.centerExpression[1] = sum / expressionSum
sum = 0.0
for key in sortedKeys[6:14]:
sum += stretch.profile[key]
alignment.centerExpression[2] = sum / expressionSum
aDC.commit(id_alignment)
def markCenterExpression(aFN, cName, rn = None, tn = None):
aNX = cgNexusFlat.Nexus(aFN, cgAlignmentFlat.cgAlignment)
aNX.load(['centerExpression', 'tTcc', 'tStart', 'sLength'], [rn, tn])
for aID in aNX.centerExpression:
aNX.centerExpression[aID] = [0.0, 0.0, 0.0]
chrom, strand, start, end = bioLibCG.tccSplit(aNX.tTcc[aID])
offset = aNX.tStart[aID]
sLen = aNX.sLength[aID]
if strand == '1':
start = start - 19 + offset
end = start + sLen
else:
end = end + 19 - offset
start = end - sLen
scanRange = bioLibCG.makeTcc(chrom, strand, start, end)
stretch = cgPeaks.stretch(scanRange, cName)
expressionSum = stretch.getSumOfLevels()
sortedKeys = stretch.profile.keys()
sortedKeys.sort()
if strand == '-1':
sortedKeys.reverse()
if expressionSum != 0:
sum = 0.0
for key in sortedKeys[8:12]:
sum += stretch.profile[key]
aNX.centerExpression[aID][0] = sum/expressionSum
sum = 0.0
for key in sortedKeys[7:13]:
sum += stretch.profile[key]
aNX.centerExpression[aID][1] = sum/expressionSum
sum = 0.0
for key in sortedKeys[6:14]:
sum += stretch.profile[key]
aNX.centerExpression[aID][2] = sum/expressionSum
aNX.save()
def markCenterExpression(aFN, cName, rn=None, tn=None):
aNX = cgNexusFlat.Nexus(aFN, cgAlignmentFlat.cgAlignment)
aNX.load(['centerExpression', 'tTcc', 'tStart', 'sLength'], [rn, tn])
for aID in aNX.centerExpression:
aNX.centerExpression[aID] = [0.0, 0.0, 0.0]
chrom, strand, start, end = bioLibCG.tccSplit(aNX.tTcc[aID])
offset = aNX.tStart[aID]
sLen = aNX.sLength[aID]
if strand == '1':
start = start - 19 + offset
end = start + sLen
else:
end = end + 19 - offset
start = end - sLen
scanRange = bioLibCG.makeTcc(chrom, strand, start, end)
stretch = cgPeaks.stretch(scanRange, cName)
expressionSum = stretch.getSumOfLevels()
sortedKeys = stretch.profile.keys()
sortedKeys.sort()
if strand == '-1':
sortedKeys.reverse()
if expressionSum != 0:
sum = 0.0
for key in sortedKeys[8:12]:
sum += stretch.profile[key]
aNX.centerExpression[aID][0] = sum / expressionSum
sum = 0.0
for key in sortedKeys[7:13]:
sum += stretch.profile[key]
aNX.centerExpression[aID][1] = sum / expressionSum
sum = 0.0
for key in sortedKeys[6:14]:
sum += stretch.profile[key]
aNX.centerExpression[aID][2] = sum / expressionSum
aNX.save()
def parallelMakePeaks(tcc, cName, minExpression):
conf = c.getConfig(cName)
f = open('out/peakData.%s.%s.%s' % (tcc, minExpression, conf.conf['assembly']), 'w')
chrom, strand, start, end = cg.tccSplit(tcc)
peaks = cgPeaks.stretch(tcc, cName)
print 'getting peaks'
peaks.createPeaks(span = 1, minVal = int(minExpression))
for x in peaks.peaks:
print x
newTcc = cg.makeTcc(chrom, strand, x, x + 1)
testedPeak = extendPeakTest(newTcc, 20, .2, .05, 0, 6, cName)
#testedPeak = roofPeakTest(newTcc, 30, .85, .9, .2, 6, 17, 24, cName)
if testedPeak:
f.write('%s\n' % testedPeak)
f.close()
def profileTargetsHisto(tccList, cName, name='boxplot'):
histDict = {} # {coord: []}
for tcc in tccList:
chrom, strand, start, end = cg.tccSplit(tcc)
#Get highest peak
tccStretch = cgPeaks.stretch(tcc, cName)
tccStretch.createPeaks(span=2)
highestCoord = tccStretch.getHighestPeak()
if highestCoord == None: continue
#profile around point
zPoint = cg.makeTcc(chrom, strand, highestCoord, end)
cProfile = svs.profileAroundPoint(zPoint, 200, cName, ratio=True)
for coord in cProfile:
try:
histDict[coord].append(cProfile[coord])
except: #quicker way to initialize
histDict[coord] = [cProfile[coord]]
plot.boxPlotHisto(histDict, name=name)
def plotProfile(tcc, cName, directory = None, min = 0):
if not directory:
fN = tcc + '.png'
else:
fN = directory + '/' + tcc + '.png'
tccStretch = cgPeaks.stretch(tcc, cName)
highest = tccStretch.getHighestLevel()
if highest < min:
return 0
sortedX = tccStretch.profile.keys()
sortedX.sort()
sortedY = []
for X in sortedX:
sortedY.append(tccStretch.profile[X])
gDevice = importr('grDevices')
gDevice.png(file=fN, width=1680, height=1050)
r.plot(sortedX, sortedY, xlab = "Coordinates", ylab = "Expression Level")
r.lines(sortedX, sortedY, type = "b")
gDevice.dev_off()
def parallelMakePeaks(tcc, cName, minExpression):
conf = c.getConfig(cName)
f = open('out/peakData.%s.%s.%s' % (tcc, minExpression, conf.conf['assembly']), 'w')
chrom, strand, start, end = cg.tccSplit(tcc)
peaks = cgPeaks.stretch(tcc, cName)
print 'getting peaks'
peaks.createPeaks(span = 1, minVal = int(minExpression))
for x in peaks.peaks:
print ""
print chrom, strand, x,
newTcc = cg.makeTcc(chrom, strand, x, x + 1)
testedPeak = extendPeakTest(newTcc, 20, .2, .05, 0, 6, cName)
#testedPeak = roofPeakTest(newTcc, 30, .85, .9, .2, 8, 16, 25, cName)
if testedPeak:
f.write('%s\n' % testedPeak)
f.close()
def profileTargetsHisto(tccList, cName, name = 'boxplot'):
histDict = {} # {coord: []}
for tcc in tccList:
chrom, strand, start, end = cg.tccSplit(tcc)
#Get highest peak
tccStretch = cgPeaks.stretch(tcc, cName)
tccStretch.createPeaks(span = 2)
highestCoord = tccStretch.getHighestPeak()
if highestCoord == None: continue
#profile around point
zPoint = cg.makeTcc(chrom, strand, highestCoord, end)
cProfile = svs.profileAroundPoint(zPoint, 200, cName, ratio = True)
for coord in cProfile:
try:
histDict[coord].append(cProfile[coord])
except: #quicker way to initialize
histDict[coord] = [cProfile[coord]]
plot.boxPlotHisto(histDict, name = name)
def parallelMakePeaks(tcc, cName, minExpression):
conf = c.getConfig(cName)
f = open('out/peakData.%s.%s.%s' % (tcc, minExpression, conf.conf['assembly']), 'w')
print 'scanning range', tcc
chrom, strand, start, end = cg.tccSplit(tcc)
peaks = cgPeaks.stretch(tcc, cName)
#print 'getting peaks'
peaks.createPeaks(span = 1, minVal = int(minExpression))
print 'len peaks', len(peaks.peaks)
endCheck = 0
for x in peaks.peaks:
print x, endCheck
'''
if x < endCheck:
print 'endChecked'
continue
'''
#scan a 30 bp range around this point and find the best roof...
pRange = 40
rTcc = cg.makeTcc(chrom, strand, x, x + 1)
#now make profile for roof...
cProfile = stepVectorScan.profileAroundPoint(rTcc, pRange, cName, ratio = True)
#now get highest stretch length and the rNext coord.
minVal = .70
highest = 0
stretch = 0
startCurrent = None
startFinal = None
endFinal = None
for i in range(1 - pRange, pRange):
print ' ', x + i, cProfile[i]
if cProfile[i] > minVal:
print ' extending stretch'
stretch += 1
if startCurrent == None:
startCurrent = i
else:
if stretch > 0:
print 'end of stretch'
if stretch > highest: #stretch ended and was higher than previous
highest = stretch
endFinal = i - 1
startFinal = startCurrent
startCurrent = None
else:
startCurrent = None
stretch = 0
#get +/- extend value...
val = [1.0, 1.0]
extend = 1
if (startFinal) and (endFinal):
low = startFinal - extend
high = endFinal + extend
if low > (1 - pRange) and high < pRange:
val[0] = float(cProfile[startFinal - extend])
val[1] = float(cProfile[endFinal + extend])
else:
print 'out of range'
continue
else:
print 'no start and end of peak'
continue
print low, high, x, endFinal
endCheck = x + endFinal
#avg expression around peak check...
#get total expression before peak
noiseExpression = 0
lowRange = range(1 - pRange, low)
highRange = range(high + 1, pRange)
totalLength = len(lowRange) + len(highRange)
for i in lowRange:
noiseExpression += cProfile[i]
for i in highRange:
noiseExpression += cProfile[i]
avgNoise = noiseExpression/float(totalLength)
#filter out peaks that look a certain way.
print highest, val[0], val[1], avgNoise
if 0 < highest < 5: #rooflength 14/26
if val[0] < 0.20 and val[1] < .20: #drop values
if avgNoise < .3:
goodTcc = cg.makeTcc(chrom, strand, x + low, x + high)
print '*KEEPER'
f.write('%s\n' % goodTcc)
f.close()
print 'DONE', tcc
def makePeakInput(cName, minExpression=2000):
mConf = c.getConfig('Main.conf')
conf = c.getConfig(cName)
assembly = conf.conf['assembly']
tccList = []
chromLens = cg.returnChromLengthDict(assembly)
f = open('peakData.%s' % minExpression, 'w')
for chrom in chromLens:
if chrom not in cg.acceptableChroms: continue
for strand in ['1', '-1']:
print 'Getting Peaks for ', chrom, strand
prevI = 0
endCheck = 0
for i in rangePoints(1, chromLens[chrom], 1000):
if i == 1:
prevI = i
continue
start = prevI
end = i
prevI = i
tcc = cg.makeTcc(chrom, strand, start, end)
#print 'scanning range', tcc
peaks = cgPeaks.stretch(tcc, cName)
peaks.createPeaks(span=3, minVal=minExpression)
for x in peaks.peaks:
if x < endCheck:
continue
#scan a 30 bp range around this point and find the best roof...
pRange = 30
rTcc = cg.makeTcc(chrom, strand, x, x + 1)
#now make profile for roof...
cProfile = stepVectorScan.profileAroundPoint(rTcc,
pRange,
cName,
ratio=True)
#now get highest stretch length and the rNext coord.
minVal = .80
highest = 0
stretch = 0
startCurrent = None
startFinal = None
endFinal = None
for i in range(1 - pRange, pRange):
if cProfile[i] > minVal:
stretch += 1
if startCurrent == None:
startCurrent = i
else:
if stretch > 0:
if stretch > highest: #stretch ended and was higher than previous
highest = stretch
endFinal = i - 1
startFinal = startCurrent
startCurrent = None
else:
startCurrent = None
stretch = 0
#get +/- 4 value...
val = [1.0, 1.0]
if (startFinal) and (endFinal):
low = startFinal - 4
high = endFinal + 4
if low > (1 - pRange) and high < pRange:
val[0] = float(cProfile[startFinal - 4])
val[1] = float(cProfile[endFinal + 4])
else:
continue
else:
continue
endCheck = x + high
#filter out peaks that look a certain way.
if 14 < highest < 26: #rooflength
if val[0] < 0.2 and val[1] < .2: #drop values
goodTcc = cg.makeTcc(chrom, strand, x + low,
x + high)
#print goodTcc
f.write('%s\n' % goodTcc)
f.close()
def findPeaks(pType, cName = None):
#init
mConf = c.cgConfig('Main.conf')
conf = c.getConfig(cName)
if pType == 'E':
predName = conf.conf['resultsExonsSorted']
else:
predName = conf.conf['resultsIntronsSorted']
print predName
#make CID:hairpin:peak dictionary
cHairs = getHairpins.getHairpins(predName)
peakDict = {}
for CID in cHairs:
peakDict[CID] = [cHairs[CID],'None']
timer = cg.cgTimer()
timer.start()
#put peaks in memory
print 'Creating peak data'
peaks = {} # chr:peak:value
for CID in cHairs:
chrom, strand, start, end = cg.tccSplit(cHairs[CID])
tcc = cHairs[CID]
#init dictionary
if chrom not in peaks:
peaks[chrom] = {}
if strand not in peaks[chrom]:
peaks[chrom][strand] = {}
#create peaks for tcc and add to peak dictionary
stretch = cgPeaks.stretch(tcc, cName)
stretch.createPeaks()
for peakCoord in stretch.peaks:
peaks[chrom][strand][peakCoord] = 0
print timer.split()
print 'finding best combos'
bestCombos = []
aPass = 0
bPass = 0
cPass = 0
numT = 0
for CID in peakDict:
cgFlag = False
if CID == '538':cgFlag = True
tcc = peakDict[CID][0]
#print tcc
tccPeaks = []
chrom = cg.ss(tcc, ':')[0]
strand = cg.ss(tcc, ':')[1]
start = int(cg.ss(tcc, ':')[2])
end = int(cg.ss(tcc, ':')[3])
#get all peaks
for i in range(start, end + 1):
if i in peaks[chrom][strand]:
#print ' peak added', i
tccPeaks.append(i)
#Calculate parameters...
pairStrings = [] #used to check if pair already added
peakCombos = []
for x in tccPeaks:
#scan a 30 bp range around this point and find the best roof...
pRange = 30
rTcc = cg.makeTcc(chrom, strand, x, x + 1)
#quickly get max value...kinda a long way to do it but whatever
cProfile = stepVectorScan.profileAroundPoint(rTcc, 1, cName, ratio = False)
xval = cProfile[0]
max = xval
highestValueCoord = x
#now make profile for roof...
cProfile = stepVectorScan.profileAroundPoint(rTcc, pRange, cName, ratio = True)
#now get highest stretch length and the rNext coord.
minVal = .80
highest = 0
stretch = 0
startCurrent = None
startFinal = None
endFinal = None
for i in range(1 - pRange, pRange):
if cProfile[i] > minVal:
stretch += 1
if startCurrent == None:
startCurrent = i
else:
if stretch > 0:
if stretch > highest: #stretch ended and was higher than previous
highest = stretch
endFinal = i - 1
startFinal = startCurrent
startCurrent = None
else:
startCurrent = None
stretch = 0
#get +/- 4 value...
val = [1.0, 1.0]
if (startFinal) and (endFinal):
low = startFinal - 4
high = endFinal + 4
if low > (1 - pRange):
if high < pRange:
val[0] = float(cProfile[startFinal - 4])
val[1] = float(cProfile[endFinal + 4])
#fill in other details...
y = 'S'
dist = 'S'
ratio = 'S'
peakCombos.append([tcc,x,y,dist,ratio,max,highest,val])
#print ' ', peakCombos[-1]
#find best combo...
topCombo = None
for combo in peakCombos:
roofLength = combo[6]
dropValue = combo[7][0]
if combo[7][1] > dropValue:
dropValue = combo[7][1]
#print roofLength, dropValue
if 14 < roofLength < 26:
if 0.0 < dropValue < 0.2:
#pick one with rooflength nearest 20:
if topCombo:
if (math.fabs(22 - roofLength)) < (math.fabs(22 - topCombo[6])):
topCombo = combo
else:
topCombo = combo
if topCombo:
peakDict[CID][1] = topCombo
bestCombos.append(topCombo)
print bestCombos[-1]
else:
#print 'None'
pass
print timer.split()
#now update predFile (SLOT 13)
predFile = open(predName, 'r')
newLines = []
for line in predFile:
CID = cg.ss(line)[7]
if peakDict[CID][1] == 'None':
peakInfo = 'None'
else:
peakInfo = '%s:%s:%s:%s:%s:%s' % (str(peakDict[CID][1][1])[-3:], 'S', str(peakDict[CID][1][4]).split('.')[0], peakDict[CID][1][5],peakDict[CID][1][6], peakDict[CID][1][7])
newLines.append(cg.appendToLine(line, peakInfo, 13))
predFile.close()
predFile = open(predName, 'w')
predFile.writelines(newLines)
predFile.close()
def markCenterExpressionOLD(smallFN, targetFN, alignmentFN, cName, outFN):
#print 'making target dict'
#make targetDict
f = open(targetFN, 'r')
targetDict = {} # tID: tLoc
for line in f:
ls = line.strip().split('\t')
targetDict[int(ls[0])] = ls[1]
f.close()
#print 'making alignment dict'
#make alignmentDict
alignDict = {} # sid: {target: offset}
f = open(alignmentFN, 'r')
for line in f:
ls = line.strip().split(' ')
sID = int(ls[0])
tID = int(ls[1])
offset = int(ls[4])
if not sID in alignDict:
alignDict[sID] = {}
alignDict[sID][tID] = offset #assumes one source to target...
f.close()
f = open(smallFN, 'r')
fOut = open(outFN, 'w')
for line in f:
ls = line.strip().split('\t')
sID = int(ls[0])
sLoc = ls[1]
sLen = len(sLoc) #This is the sequence for simulated reads...
#sLen = bioLibCG.getTccLength(sLoc) #off by one?
tIDs = ls[4].split(',')
for tID in tIDs:
tID = int(tID)
tLoc = targetDict[tID]
chrom, strand, start, end = bioLibCG.tccSplit(tLoc)
offset = alignDict[sID][tID]
if strand == '1':
start = start - 19 + offset
end = start + sLen
else:
end = end + 19 - offset
start = end - sLen
scanRange = bioLibCG.makeTcc(chrom, strand, start, end)
stretch = cgPeaks.stretch(scanRange, cName)
expressionSum = stretch.getSumOfLevels()
sortedKeys = stretch.profile.keys()
sortedKeys.sort()
if strand == '-1':
sortedKeys.reverse()
lowE = 0.0
midE = 0.0
highE = 0.0
if expressionSum != 0:
sum = 0.0
for key in sortedKeys[8:12]:
sum += stretch.profile[key]
lowE = sum / expressionSum
sum = 0.0
for key in sortedKeys[7:13]:
sum += stretch.profile[key]
midE = sum / expressionSum
sum = 0.0
for key in sortedKeys[6:14]:
sum += stretch.profile[key]
highE = sum / expressionSum
fOut.write('%s\t%s\t%s\t%s\t%s\n' % (sID, tID, lowE, midE, highE))
import math
import bioLibCG as cg
knowns = compare.tccFileToList("mouseKnownMirs.tcc", 0)
eLevels = []
for known in knowns:
chrom, strand, start, end = cg.tccSplit(known, True) # text...
if strand == "1":
strand = "-1"
else:
strand = "1"
oppTcc = cg.makeTcc(chrom, strand, start, end)
knownStretch = cgPeaks.stretch(known)
knownStretch.createPeaks(1, 20)
kPos = knownStretch.getHighestPeak()
if kPos:
eLevels.append(knownStretch.profile[kPos])
oppStretch = cgPeaks.stretch(oppTcc)
oppStretch.createPeaks(1, 20)
oPos = oppStretch.getHighestPeak()
if oPos and kPos:
# determine if they are close enough to be considered mirrored...
if math.fabs(int(kPos) - int(oPos)) < 12:
print known, oPos, kPos, oppStretch.profile[oPos], knownStretch.profile[kPos]
def makeFigure(fN, targetFN, alignmentFN, cName):
# make targetDict
f = open(targetFN, "r")
targetDict = {} # tID: tLoc
for line in f:
ls = line.strip().split("\t")
targetDict[int(ls[0])] = ls[1]
f.close()
# make alignmentDict
alignDict = {} # sid: {target: offset}
f = open(alignmentFN, "r")
for line in f:
ls = line.strip().split(" ")
sID = int(ls[0])
tID = int(ls[1])
offset = int(ls[4])
if not sID in alignDict:
alignDict[sID] = {}
alignDict[sID][tID] = offset # assumes one source to target...
f.close()
f = open(fN, "r")
histoVals = []
for line in f:
ls = line.strip().split("\t")
sID = int(ls[0])
sLoc = ls[1]
sChrom, sStrand, sStart, sEnd = bioLibCG.tccSplit(sLoc)
sLen = sEnd - sStart
tIDs = ls[4].split(",")
for tID in tIDs:
tID = int(tID)
tLoc = targetDict[tID]
chrom, strand, start, end = bioLibCG.tccSplit(tLoc)
offset = alignDict[sID][tID]
if sStrand == "1":
start = start - 19 + offset
end = start + sLen
else:
end = end + 19 - offset
start = end - sLen
scanRange = bioLibCG.makeTcc(chrom, strand, start, end)
stretch = cgPeaks.stretch(scanRange, cName)
highest = stretch.getHighestLevel()
sortedKeys = stretch.profile.keys()
if sStrand == "-1":
sortedKeys.reverse()
i = 0
for key in sortedKeys:
level = stretch.profile[key]
for j in range(0, level):
histoVals.append(i)
i += 1
cgPlot.plotHistogram(histoVals)
def updateReadDensity(tType, cName):
#go through wig each chromosome and check the mature seqs
mainConf = cgConfig.cgConfig('Main.conf')
conf = cgConfig.getConfig(cName)
organism = conf.conf['organism']
wigFolder = mainConf.conf['wig%s' % organism]
newLines = []
#Differentiate between exon or intron...
if tType == 'E':
pFileName = conf.conf['resultsExons']
elif tType == 'I':
pFileName = conf.conf['resultsIntrons']
else:
print 'READ UPDATE FAIL'
print ' Updating Read Density:', tType
#get read density for each line...
print ' calculating hits for mature seqs'
#calculate total hits per mature
mirFile = open(pFileName, 'r')
for line in mirFile:
mTcc = line.strip().split('\t')[1]
mirID = line.strip().split('\t')[0]
tccStretch = cgPeaks.stretch(mTcc, cName)
highestHit = 0
for i in range(int(mTcc.split(':')[2]), int(mTcc.split(':')[3])):
if i in tccStretch.profile:
if tccStretch.profile[i] > highestHit:
highestHit = tccStretch.profile[i]
newLines.append(cg.appendToLine(line, str(highestHit), 11))
mirFile.close()
print 'Writing New File'
#write new results file
outFile = open(pFileName, 'w')
for line in newLines:
outFile.write(line)
outFile.close()
####NOW UPDATE HIGHEST HIT PER CLUSTER####
clusterCount = {}
pFile = open(pFileName, 'r')
for line in pFile:
predictionCount = int(line.strip().split('\t')[11])
CID = line.strip().split('\t')[7]
if CID in clusterCount:
if clusterCount[CID] < predictionCount:
clusterCount[CID] = predictionCount
else:
clusterCount[CID] = predictionCount
pFile.close()
#update the file --> cluster small count
newLines = []
predFile = open(pFileName, 'r')
for line in predFile:
CID = line.strip().split('\t')[7]
numMax = clusterCount[CID]
newLines.append(cg.appendToLine(line, str(numMax), 12))
predFile.close()
#sort newLines by clusterID
sortDict = {}
CIDs = []
for line in newLines:
CID = int(line.strip().split('\t')[7])
if CID not in CIDs:
CIDs.append(CID)
if CID in sortDict:
sortDict[CID].append(line)
else:
sortDict[CID] = [line]
CIDs.sort()
newLines = []
for CID in CIDs:
for line in sortDict[CID]:
newLines.append(line)
#write new File
newFile = open(pFileName, 'w')
for line in newLines:
newFile.write(line)
newFile.close()
def makePeakInput(cName, minExpression = 2000):
mConf = c.getConfig('Main.conf')
conf = c.getConfig(cName)
assembly = conf.conf['assembly']
tccList = []
chromLens = cg.returnChromLengthDict(assembly)
f = open('peakData.%s' % minExpression, 'w')
for chrom in chromLens:
if chrom not in cg.acceptableChroms: continue
for strand in ['1', '-1']:
print 'Getting Peaks for ', chrom, strand
prevI = 0
endCheck = 0
for i in rangePoints(1, chromLens[chrom], 1000):
if i == 1:
prevI = i
continue
start = prevI
end = i
prevI = i
tcc = cg.makeTcc(chrom, strand, start, end)
#print 'scanning range', tcc
peaks = cgPeaks.stretch(tcc, cName)
peaks.createPeaks(span = 3, minVal = minExpression)
for x in peaks.peaks:
if x < endCheck:
continue
#scan a 30 bp range around this point and find the best roof...
pRange = 30
rTcc = cg.makeTcc(chrom, strand, x, x + 1)
#now make profile for roof...
cProfile = stepVectorScan.profileAroundPoint(rTcc, pRange, cName, ratio = True)
#now get highest stretch length and the rNext coord.
minVal = .80
highest = 0
stretch = 0
startCurrent = None
startFinal = None
endFinal = None
for i in range(1 - pRange, pRange):
if cProfile[i] > minVal:
stretch += 1
if startCurrent == None:
startCurrent = i
else:
if stretch > 0:
if stretch > highest: #stretch ended and was higher than previous
highest = stretch
endFinal = i - 1
startFinal = startCurrent
startCurrent = None
else:
startCurrent = None
stretch = 0
#get +/- 4 value...
val = [1.0, 1.0]
if (startFinal) and (endFinal):
low = startFinal - 4
high = endFinal + 4
if low > (1 - pRange) and high < pRange:
val[0] = float(cProfile[startFinal - 4])
val[1] = float(cProfile[endFinal + 4])
else:
continue
else:
continue
endCheck = x + high
#filter out peaks that look a certain way.
if 14 < highest < 26: #rooflength
if val[0] < 0.2 and val[1] < .2: #drop values
goodTcc = cg.makeTcc(chrom, strand, x + low, x + high)
#print goodTcc
f.write('%s\n' % goodTcc)
f.close()
def markCenterExpressionOLD(smallFN, targetFN, alignmentFN, cName, outFN):
#print 'making target dict'
#make targetDict
f = open(targetFN, 'r')
targetDict = {} # tID: tLoc
for line in f:
ls = line.strip().split('\t')
targetDict[int(ls[0])] = ls[1]
f.close()
#print 'making alignment dict'
#make alignmentDict
alignDict = {} # sid: {target: offset}
f = open(alignmentFN, 'r')
for line in f:
ls = line.strip().split(' ')
sID = int(ls[0])
tID = int(ls[1])
offset = int(ls[4])
if not sID in alignDict:
alignDict[sID] = {}
alignDict[sID][tID] = offset #assumes one source to target...
f.close()
f = open(smallFN, 'r')
fOut = open(outFN, 'w')
for line in f:
ls = line.strip().split('\t')
sID = int(ls[0])
sLoc = ls[1]
sLen = len(sLoc) #This is the sequence for simulated reads...
#sLen = bioLibCG.getTccLength(sLoc) #off by one?
tIDs = ls[4].split(',')
for tID in tIDs:
tID = int(tID)
tLoc = targetDict[tID]
chrom, strand, start, end = bioLibCG.tccSplit(tLoc)
offset = alignDict[sID][tID]
if strand == '1':
start = start - 19 + offset
end = start + sLen
else:
end = end + 19 - offset
start = end - sLen
scanRange = bioLibCG.makeTcc(chrom, strand, start, end)
stretch = cgPeaks.stretch(scanRange, cName)
expressionSum = stretch.getSumOfLevels()
sortedKeys = stretch.profile.keys()
sortedKeys.sort()
if strand == '-1':
sortedKeys.reverse()
lowE = 0.0
midE = 0.0
highE = 0.0
if expressionSum != 0:
sum = 0.0
for key in sortedKeys[8:12]:
sum += stretch.profile[key]
lowE = sum/expressionSum
sum = 0.0
for key in sortedKeys[7:13]:
sum += stretch.profile[key]
midE = sum/expressionSum
sum = 0.0
for key in sortedKeys[6:14]:
sum += stretch.profile[key]
highE = sum/expressionSum
fOut.write('%s\t%s\t%s\t%s\t%s\n' % (sID, tID, lowE, midE, highE))
def updateReadDensity(tType, cName):
#go through wig each chromosome and check the mature seqs
mainConf = cgConfig.cgConfig('Main.conf')
conf = cgConfig.getConfig(cName)
organism = conf.conf['organism']
wigFolder = mainConf.conf['wig%s' % organism]
newLines = []
#Differentiate between exon or intron...
if tType == 'E':
pFileName = conf.conf['resultsExons']
elif tType == 'I':
pFileName = conf.conf['resultsIntrons']
else:
print 'READ UPDATE FAIL'
print ' Updating Read Density:', tType
#get read density for each line...
print ' calculating hits for mature seqs'
#calculate total hits per mature
mirFile = open(pFileName, 'r')
for line in mirFile:
mTcc = line.strip().split('\t')[1]
mirID = line.strip().split('\t')[0]
tccStretch = cgPeaks.stretch(mTcc, cName)
highestHit = 0
for i in range(int(mTcc.split(':')[2]), int(mTcc.split(':')[3])):
if i in tccStretch.profile:
if tccStretch.profile[i] > highestHit:
highestHit = tccStretch.profile[i]
newLines.append(cg.appendToLine(line, str(highestHit), 11))
mirFile.close()
print 'Writing New File'
#write new results file
outFile = open(pFileName, 'w')
for line in newLines:
outFile.write(line)
outFile.close()
####NOW UPDATE HIGHEST HIT PER CLUSTER####
clusterCount = {}
pFile = open(pFileName, 'r')
for line in pFile:
predictionCount = int(line.strip().split('\t')[11])
CID = line.strip().split('\t')[7]
if CID in clusterCount:
if clusterCount[CID] < predictionCount:
clusterCount[CID] = predictionCount
else:
clusterCount[CID] = predictionCount
pFile.close()
#update the file --> cluster small count
newLines = []
predFile = open(pFileName, 'r')
for line in predFile:
CID = line.strip().split('\t')[7]
numMax = clusterCount[CID]
newLines.append(cg.appendToLine(line, str(numMax), 12))
predFile.close()
#sort newLines by clusterID
sortDict = {}
CIDs = []
for line in newLines:
CID = int(line.strip().split('\t')[7])
if CID not in CIDs:
CIDs.append(CID)
if CID in sortDict:
sortDict[CID].append(line)
else:
sortDict[CID] = [line]
CIDs.sort()
newLines = []
for CID in CIDs:
for line in sortDict[CID]:
newLines.append(line)
#write new File
newFile = open(pFileName, 'w')
for line in newLines:
newFile.write(line)
newFile.close()
import math
import bioLibCG as cg
knowns = compare.tccFileToList('mouseKnownMirs.tcc', 0)
eLevels = []
for known in knowns:
chrom, strand, start, end = cg.tccSplit(known, True) #text...
if strand == '1':
strand = '-1'
else:
strand = '1'
oppTcc = cg.makeTcc(chrom, strand, start, end)
knownStretch = cgPeaks.stretch(known)
knownStretch.createPeaks(1, 20)
kPos = knownStretch.getHighestPeak()
if kPos: eLevels.append(knownStretch.profile[kPos])
oppStretch = cgPeaks.stretch(oppTcc)
oppStretch.createPeaks(1, 20)
oPos = oppStretch.getHighestPeak()
if oPos and kPos:
#determine if they are close enough to be considered mirrored...
if math.fabs(int(kPos) - int(oPos)) < 12:
print known, oPos, kPos, oppStretch.profile[
oPos], knownStretch.profile[kPos]
print eLevels
def parallelMakePeaks(tcc, cName, minExpression):
conf = c.getConfig(cName)
f = open(
'out/peakData.%s.%s.%s' % (tcc, minExpression, conf.conf['assembly']),
'w')
print 'scanning range', tcc
chrom, strand, start, end = cg.tccSplit(tcc)
peaks = cgPeaks.stretch(tcc, cName)
#print 'getting peaks'
peaks.createPeaks(span=1, minVal=int(minExpression))
print 'len peaks', len(peaks.peaks)
endCheck = 0
for x in peaks.peaks:
print x, endCheck
'''
if x < endCheck:
print 'endChecked'
continue
'''
#scan a 30 bp range around this point and find the best roof...
pRange = 40
rTcc = cg.makeTcc(chrom, strand, x, x + 1)
#now make profile for roof...
cProfile = stepVectorScan.profileAroundPoint(rTcc,
pRange,
cName,
ratio=True)
#now get highest stretch length and the rNext coord.
minVal = .70
highest = 0
stretch = 0
startCurrent = None
startFinal = None
endFinal = None
for i in range(1 - pRange, pRange):
print ' ', x + i, cProfile[i]
if cProfile[i] > minVal:
print ' extending stretch'
stretch += 1
if startCurrent == None:
startCurrent = i
else:
if stretch > 0:
print 'end of stretch'
if stretch > highest: #stretch ended and was higher than previous
highest = stretch
endFinal = i - 1
startFinal = startCurrent
startCurrent = None
else:
startCurrent = None
stretch = 0
#get +/- extend value...
val = [1.0, 1.0]
extend = 1
if (startFinal) and (endFinal):
low = startFinal - extend
high = endFinal + extend
if low > (1 - pRange) and high < pRange:
val[0] = float(cProfile[startFinal - extend])
val[1] = float(cProfile[endFinal + extend])
else:
print 'out of range'
continue
else:
print 'no start and end of peak'
continue
print low, high, x, endFinal
endCheck = x + endFinal
#avg expression around peak check...
#get total expression before peak
noiseExpression = 0
lowRange = range(1 - pRange, low)
highRange = range(high + 1, pRange)
totalLength = len(lowRange) + len(highRange)
for i in lowRange:
noiseExpression += cProfile[i]
for i in highRange:
noiseExpression += cProfile[i]
avgNoise = noiseExpression / float(totalLength)
#filter out peaks that look a certain way.
print highest, val[0], val[1], avgNoise
if 0 < highest < 5: #rooflength 14/26
if val[0] < 0.20 and val[1] < .20: #drop values
if avgNoise < .3:
goodTcc = cg.makeTcc(chrom, strand, x + low, x + high)
print '*KEEPER'
f.write('%s\n' % goodTcc)
f.close()
print 'DONE', tcc
def makeFigure(fN, targetFN, alignmentFN, cName):
#make targetDict
f = open(targetFN, 'r')
targetDict = {} # tID: tLoc
for line in f:
ls = line.strip().split('\t')
targetDict[int(ls[0])] = ls[1]
f.close()
#make alignmentDict
alignDict = {} # sid: {target: offset}
f = open(alignmentFN, 'r')
for line in f:
ls = line.strip().split(' ')
sID = int(ls[0])
tID = int(ls[1])
offset = int(ls[4])
if not sID in alignDict:
alignDict[sID] = {}
alignDict[sID][tID] = offset #assumes one source to target...
f.close()
f = open(fN, 'r')
histoVals = []
for line in f:
ls = line.strip().split('\t')
sID = int(ls[0])
sLoc = ls[1]
sChrom, sStrand, sStart, sEnd = bioLibCG.tccSplit(sLoc)
sLen = sEnd - sStart
tIDs = ls[4].split(',')
for tID in tIDs:
tID = int(tID)
tLoc = targetDict[tID]
chrom, strand, start, end = bioLibCG.tccSplit(tLoc)
offset = alignDict[sID][tID]
if sStrand == '1':
start = start - 19 + offset
end = start + sLen
else:
end = end + 19 - offset
start = end - sLen
scanRange = bioLibCG.makeTcc(chrom, strand, start, end)
stretch = cgPeaks.stretch(scanRange, cName)
highest = stretch.getHighestLevel()
sortedKeys = stretch.profile.keys()
if sStrand == '-1':
sortedKeys.reverse()
i = 0
for key in sortedKeys:
level = stretch.profile[key]
for j in range(0,level):
histoVals.append(i)
i += 1
cgPlot.plotHistogram(histoVals)
def findPeaks(pType, cName=None):
#init
mConf = c.cgConfig('Main.conf')
conf = c.getConfig(cName)
if pType == 'E':
predName = conf.conf['resultsExonsSorted']
else:
predName = conf.conf['resultsIntronsSorted']
print predName
#make CID:hairpin:peak dictionary
cHairs = getHairpins.getHairpins(predName)
peakDict = {}
for CID in cHairs:
peakDict[CID] = [cHairs[CID], 'None']
timer = cg.cgTimer()
timer.start()
#put peaks in memory
print 'Creating peak data'
peaks = {} # chr:peak:value
for CID in cHairs:
chrom, strand, start, end = cg.tccSplit(cHairs[CID])
tcc = cHairs[CID]
#init dictionary
if chrom not in peaks:
peaks[chrom] = {}
if strand not in peaks[chrom]:
peaks[chrom][strand] = {}
#create peaks for tcc and add to peak dictionary
stretch = cgPeaks.stretch(tcc, cName)
stretch.createPeaks()
for peakCoord in stretch.peaks:
peaks[chrom][strand][peakCoord] = 0
print timer.split()
print 'finding best combos'
bestCombos = []
aPass = 0
bPass = 0
cPass = 0
numT = 0
for CID in peakDict:
cgFlag = False
if CID == '538': cgFlag = True
tcc = peakDict[CID][0]
#print tcc
tccPeaks = []
chrom = cg.ss(tcc, ':')[0]
strand = cg.ss(tcc, ':')[1]
start = int(cg.ss(tcc, ':')[2])
end = int(cg.ss(tcc, ':')[3])
#get all peaks
for i in range(start, end + 1):
if i in peaks[chrom][strand]:
#print ' peak added', i
tccPeaks.append(i)
#Calculate parameters...
pairStrings = [] #used to check if pair already added
peakCombos = []
for x in tccPeaks:
#scan a 30 bp range around this point and find the best roof...
pRange = 30
rTcc = cg.makeTcc(chrom, strand, x, x + 1)
#quickly get max value...kinda a long way to do it but whatever
cProfile = stepVectorScan.profileAroundPoint(rTcc,
1,
cName,
ratio=False)
xval = cProfile[0]
max = xval
highestValueCoord = x
#now make profile for roof...
cProfile = stepVectorScan.profileAroundPoint(rTcc,
pRange,
cName,
ratio=True)
#now get highest stretch length and the rNext coord.
minVal = .80
highest = 0
stretch = 0
startCurrent = None
startFinal = None
endFinal = None
for i in range(1 - pRange, pRange):
if cProfile[i] > minVal:
stretch += 1
if startCurrent == None:
startCurrent = i
else:
if stretch > 0:
if stretch > highest: #stretch ended and was higher than previous
highest = stretch
endFinal = i - 1
startFinal = startCurrent
startCurrent = None
else:
startCurrent = None
stretch = 0
#get +/- 4 value...
val = [1.0, 1.0]
if (startFinal) and (endFinal):
low = startFinal - 4
high = endFinal + 4
if low > (1 - pRange):
if high < pRange:
val[0] = float(cProfile[startFinal - 4])
val[1] = float(cProfile[endFinal + 4])
#fill in other details...
y = 'S'
dist = 'S'
ratio = 'S'
peakCombos.append([tcc, x, y, dist, ratio, max, highest, val])
#print ' ', peakCombos[-1]
#find best combo...
topCombo = None
for combo in peakCombos:
roofLength = combo[6]
dropValue = combo[7][0]
if combo[7][1] > dropValue:
dropValue = combo[7][1]
#print roofLength, dropValue
if 14 < roofLength < 26:
if 0.0 < dropValue < 0.2:
#pick one with rooflength nearest 20:
if topCombo:
if (math.fabs(22 - roofLength)) < (
math.fabs(22 - topCombo[6])):
topCombo = combo
else:
topCombo = combo
if topCombo:
peakDict[CID][1] = topCombo
bestCombos.append(topCombo)
print bestCombos[-1]
else:
#print 'None'
pass
print timer.split()
#now update predFile (SLOT 13)
predFile = open(predName, 'r')
newLines = []
for line in predFile:
CID = cg.ss(line)[7]
if peakDict[CID][1] == 'None':
peakInfo = 'None'
else:
peakInfo = '%s:%s:%s:%s:%s:%s' % (
str(peakDict[CID][1][1])[-3:], 'S', str(
peakDict[CID][1][4]).split('.')[0], peakDict[CID][1][5],
peakDict[CID][1][6], peakDict[CID][1][7])
newLines.append(cg.appendToLine(line, peakInfo, 13))
predFile.close()
predFile = open(predName, 'w')
predFile.writelines(newLines)
predFile.close()
