def generateReference(self):
lrepeat = self.repeatLen
startLoc = self.G / 2
endLoc = startLoc + lrepeat
locListTmp = dataGenLib.locList()
repeatInfoTmp = dataGenLib.repeatInfo()
for i in range(self.m):
repeatInfoTmp.addRepeatCopy(i, startLoc, endLoc)
locListTmp.addRepeatInfo(repeatInfoTmp)
seqList = []
for i in range(self.m):
seqList.append(dataGenLib.randomStringGen(self.G))
seqList = dataGenLib.insertRepeat(seqList, locListTmp)
newSeqList = ["", ""]
pt1, pt2 = startLoc + lrepeat / 3, startLoc + 2 * lrepeat / 3
print pt1, pt2
newSeqList[0] = seqList[0][0:pt1] + 'A' + seqList[0][
pt1 + 1:pt2] + 'G' + seqList[0][pt2 + 1:]
newSeqList[1] = seqList[1][0:pt1] + 'C' + seqList[1][
pt1 + 1:pt2] + 'T' + seqList[1][pt2 + 1:]
dataGenLib.writeListToFile(self.folderName, 'reference.fasta',
newSeqList)
def genearteReads(self):
genList = dataGenLib.readFromFasta(self.folderName + "reference.fasta")
NList = []
segList = []
for c in self.abun:
NList.append(int(self.G * c / self.L))
for i in range(len(genList)):
for j in range(NList[i]):
segList.append(
dataGenLib.createANoisyRead(self.L, self.p, genList[i]))
dataGenLib.writeListToFile(self.folderName, "raw_reads.fasta", segList)
def generateContigs(self):
genList = dataGenLib.readFromFasta(self.folderName + "reference.fasta")
breakPoints = []
numCopy = 3
separation = 10**6
for i in range(1, 1 + numCopy):
breakPoints.append([0, i * separation])
segList = dataGenLib.contigBreakDown(genList, breakPoints)
seqListNew = [segList[0] + segList[2] + segList[1] + segList[3]]
dataGenLib.writeListToFile(self.folderName, 'contigs.fasta',
seqListNew)
def generateContigs(self):
print "Generate contigs"
genList = dataGenLib.readFromFasta(self.folderName + "reference.fasta")
breakPoints = []
for i in range(len(genList)):
G = len(genList[i])
breakPoints.append([i, G / 2])
segList = dataGenLib.contigBreakDown(genList, breakPoints)
seqListNew = [segList[0] + segList[3], segList[2] + segList[1]]
for eachitem in seqListNew:
print len(eachitem)
dataGenLib.writeListToFile(self.folderName, 'contigs.fasta',
seqListNew)
def generateReference(self):
lrepeat = 1000
numCopy = 3
separation = 10**6
locListTmp = dataGenLib.locList()
repeatInfoTmp = dataGenLib.repeatInfo()
for i in range(1, 1 + numCopy):
repeatInfoTmp.addRepeatCopy(0, i * separation,
i * separation + lrepeat)
locListTmp.addRepeatInfo(repeatInfoTmp)
seqList = [dataGenLib.randomStringGen(self.G)]
seqList = dataGenLib.insertRepeat(seqList, locListTmp)
dataGenLib.writeListToFile(self.folderName, 'reference.fasta', seqList)
def generateReference(self):
lrepeat = 5000
startLoc = self.G / 2
endLoc = startLoc + lrepeat
locListTmp = dataGenLib.locList()
repeatInfoTmp = dataGenLib.repeatInfo()
for i in range(self.m):
repeatInfoTmp.addRepeatCopy(i, startLoc, endLoc)
locListTmp.addRepeatInfo(repeatInfoTmp)
seqList = []
for i in range(self.m):
seqList.append(dataGenLib.randomStringGen(self.G))
seqList = dataGenLib.insertRepeat(seqList, locListTmp)
dataGenLib.writeListToFile(self.folderName, 'reference.fasta', seqList)