def replaceFiles(folderName, replacedName):
commandList = []
commandList.append("cp " + folderName + "improved3.fasta " + folderName +
"improved3_backup.fasta")
commandList.append("cp " + folderName + "improved3_Double.fasta " +
folderName + "improved3_backup.fasta")
IORobot.writeToFile_Double1(folderName, replacedName[0:-6] + ".fasta",
replacedName[0:-6] + "_Double.fasta", "contig")
commandList.append("cp " + folderName + replacedName + " " + folderName +
"improved3.fasta")
command = "perl -pe 's/>[^\$]*$/\">Segkk\" . $n++ .\"\n\"/ge' " + folderName + "improved3.fasta > " + folderName + "newImproved3.fasta "
commandList.append(command)
command = "cp " + folderName + "newImproved3.fasta " + folderName + "improved3.fasta "
commandList.append(command)
commandList.append("cp " + folderName + replacedName[0:-6] +
"_Double.fasta " + folderName +
"improved3_Double.fasta")
for eachcommand in commandList:
print eachcommand
os.system(eachcommand)
def replaceFiles( folderName, replacedName) :
commandList = []
commandList.append("cp " + folderName + "improved3.fasta " + folderName + "improved3_backup.fasta")
commandList.append("cp " + folderName + "improved3_Double.fasta " + folderName + "improved3_backup.fasta")
IORobot.writeToFile_Double1(folderName, replacedName[0:-6]+".fasta", replacedName[0:-6]+"_Double.fasta", "contig")
commandList.append("cp " + folderName + replacedName + " "+folderName + "improved3.fasta")
command = "perl -pe 's/>[^\$]*$/\">Segkk\" . $n++ .\"\n\"/ge' "+folderName+"improved3.fasta > "+folderName+"newImproved3.fasta "
commandList.append(command)
command = "cp " +folderName+"newImproved3.fasta "+folderName+"improved3.fasta "
commandList.append(command)
commandList.append("cp " + folderName + replacedName[0:-6]+"_Double.fasta " + folderName + "improved3_Double.fasta")
for eachcommand in commandList:
print eachcommand
os.system(eachcommand)
def formReadContigStringGraph(folderName,
mummerLink,
contigFilename,
readsetFilename,
optTypeFileHeader,
graphName,
needAlignment=True):
'''
Input : all_associated_reads.fasta, improved3.fasta
Output : (G) String Graph linking the reads and contigs
Algorithm:
a) Form double reads and contigs V
b) Mummer the data and extract dataList three times V
c) Use the subroutine to output a graph V
d) Output the graph to a file phasing_String_graph.graph V
'''
G = []
IORobot.writeToFile_Double1(folderName, contigFilename + ".fasta",
contigFilename + "_Double.fasta", "contig")
IORobot.writeToFile_Double1(folderName, readsetFilename + ".fasta",
readsetFilename + "_Double.fasta", "reads")
header, referenceFile, queryFile = optTypeFileHeader + "CC", contigFilename + "_Double.fasta", contigFilename + "_Double.fasta"
#if needAlignment:
# alignerRobot.useMummerAlign(mummerLink, folderName, header, referenceFile, queryFile)
if needAlignment:
alignerRobot.useMummerAlignBatch(
mummerLink, folderName, [[header, referenceFile, queryFile, ""]],
houseKeeper.globalParallel)
lenDicCC = IORobot.obtainLength(folderName,
contigFilename + "_Double.fasta")
dataListCC = alignerRobot.extractMumData(folderName, header + "Out")
dataListCC = abunHouseKeeper.filterData(dataListCC, lenDicCC)
header, referenceFile, queryFile = optTypeFileHeader + "RR", readsetFilename + "_Double.fasta", readsetFilename + "_Double.fasta"
lenDicRR = IORobot.obtainLength(folderName,
readsetFilename + "_Double.fasta")
if not abunHouseKeeper.abunGlobalRRDisable:
if needAlignment:
alignerSubRoutine(folderName, referenceFile, queryFile, mummerLink,
header)
dataListRR = alignerRobot.extractMumData(folderName, header + "Out")
dataListRR = abunHouseKeeper.filterData(dataListRR, lenDicRR)
dataListRR = abunHouseKeeper.filterDataIdentical(dataListRR, lenDicRR)
else:
dataListRR = []
header, referenceFile, queryFile = optTypeFileHeader + "CR", contigFilename + "_Double.fasta", readsetFilename + "_Double.fasta"
if needAlignment:
alignerSubRoutine(folderName, referenceFile, queryFile, mummerLink,
header)
lenDicCR = dict(lenDicCC.items() + lenDicRR.items())
dataListCR = alignerRobot.extractMumData(folderName, header + "Out")
dataListCR = abunHouseKeeper.filterData(dataListCR, lenDicCR)
numberOfNodes = len(lenDicCR)
G = graphLib.seqGraph(numberOfNodes)
N1, N2 = len(lenDicCC), len(lenDicRR)
print "N1, N2, numberOfNodes: ", N1, N2, numberOfNodes
'''
e.g. of dataListCC[0], dataListRR[0], dataListCR[0]
[1, 520, 2913194, 2913716, 520, 523, 99.05, 'Contig0_d', 'Contig2_d']
[1, 1383, 1253, 2603, 1383, 1351, 82.39, 'Read0_d', 'Read1705_p']
[1, 718, 4334, 5074, 718, 741, 91.91, 'Contig0_d', 'Read1018_d']
'''
addDataToList(dataListCC, G, 0, 0, 'C', 'C')
addDataToList(dataListRR, G, N1, N1, 'R', 'R')
addDataToList(dataListCR, G, 0, N1, 'C', 'R')
Gnew = formExtraEdges(folderName, optTypeFileHeader, contigFilename, G, N1)
Gnew.saveToFile(folderName, graphName)
print "len(Gnew.graphNodesList)", len(Gnew.graphNodesList)
def formReadContigStringGraph(folderName, mummerLink, contigFilename, readsetFilename, optTypeFileHeader, graphName):
'''
Input : all_associated_reads.fasta, improved3.fasta
Output : (G) String Graph linking the reads and contigs
Algorithm:
a) Form double reads and contigs V
b) Mummer the data and extract dataList three times V
c) Use the subroutine to output a graph V
d) Output the graph to a file phasing_String_graph.graph V
'''
G = []
IORobot.writeToFile_Double1(folderName, contigFilename + ".fasta", contigFilename + "_Double.fasta", "contig")
IORobot.writeToFile_Double1(folderName, readsetFilename + ".fasta", readsetFilename + "_Double.fasta", "reads")
header, referenceFile, queryFile = optTypeFileHeader + "CC", contigFilename + "_Double.fasta" , contigFilename + "_Double.fasta"
if True:
alignerRobot.useMummerAlign(mummerLink, folderName, header, referenceFile, queryFile)
lenDicCC = IORobot.obtainLength(folderName, contigFilename + "_Double.fasta")
dataListCC = alignerRobot.extractMumData(folderName, header + "Out")
dataListCC = abunHouseKeeper.filterData(dataListCC, lenDicCC)
header, referenceFile, queryFile = optTypeFileHeader + "RR", readsetFilename + "_Double.fasta" , readsetFilename + "_Double.fasta"
lenDicRR = IORobot.obtainLength(folderName, readsetFilename + "_Double.fasta")
if not abunHouseKeeper.abunGlobalRRDisable:
if True:
alignerSubRoutine(folderName ,referenceFile, queryFile, mummerLink, header )
dataListRR = alignerRobot.extractMumData(folderName, header + "Out")
dataListRR = abunHouseKeeper.filterData(dataListRR, lenDicRR)
for eachitem in dataListRR:
if eachitem[-2] == "Read164_p" and eachitem[-1] == "Read159_p" :
print "debug" , eachitem
if eachitem[-1] == "Read164_p" and eachitem[-2] == "Read159_p" :
print "debug" , eachitem
dataListRR = abunHouseKeeper.filterDataIdentical(dataListRR, lenDicRR)
else:
dataListRR = []
header, referenceFile, queryFile = optTypeFileHeader + "CR", contigFilename + "_Double.fasta" , readsetFilename + "_Double.fasta"
if True:
alignerSubRoutine(folderName ,referenceFile, queryFile, mummerLink, header )
#alignerRobot.useMummerAlign(mummerLink, folderName, header, referenceFile, queryFile)
lenDicCR = dict(lenDicCC.items() + lenDicRR.items())
dataListCR = alignerRobot.extractMumData(folderName, header + "Out")
dataListCR = abunHouseKeeper.filterData(dataListCR, lenDicCR)
numberOfNodes = len(lenDicCR)
G = graphLib.seqGraph(numberOfNodes)
N1, N2 = len(lenDicCC), len(lenDicRR)
print "N1, N2, numberOfNodes: ", N1, N2, numberOfNodes
'''
e.g. of dataListCC[0], dataListRR[0], dataListCR[0]
[1, 520, 2913194, 2913716, 520, 523, 99.05, 'Contig0_d', 'Contig2_d']
[1, 1383, 1253, 2603, 1383, 1351, 82.39, 'Read0_d', 'Read1705_p']
[1, 718, 4334, 5074, 718, 741, 91.91, 'Contig0_d', 'Read1018_d']
'''
# print dataListCC[0]
# print dataListRR[0]
# print dataListCR[0]
# for eachitem in dataListCC:
# print eachitem
addDataToList(dataListCC, G, 0, 0, 'C', 'C')
# for eachitem in dataListRR[0:10]:
# print eachitem , lenDicRR[eachitem[-2]], lenDicRR[eachitem[-1]]
addDataToList(dataListRR, G, N1, N1, 'R', 'R')
addDataToList(dataListCR, G, 0, N1, 'C', 'R')
# G.reportEdge()
G.saveToFile(folderName, graphName)
checkGraphLength(G, N1, lenDicRR)
# print len(G.graphNodesList[0].listOfPrevNodes), len(G.graphNodesList[0].listOfNextNodes)
print "len(G.graphNodesList)", len(G.graphNodesList)
def formReadContigStringGraph(folderName, mummerLink, contigFilename,
readsetFilename, optTypeFileHeader, graphName):
'''
Input : all_associated_reads.fasta, improved3.fasta
Output : (G) String Graph linking the reads and contigs
Algorithm:
a) Form double reads and contigs V
b) Mummer the data and extract dataList three times V
c) Use the subroutine to output a graph V
d) Output the graph to a file phasing_String_graph.graph V
'''
G = []
IORobot.writeToFile_Double1(folderName, contigFilename + ".fasta",
contigFilename + "_Double.fasta", "contig")
IORobot.writeToFile_Double1(folderName, readsetFilename + ".fasta",
readsetFilename + "_Double.fasta", "reads")
header, referenceFile, queryFile = optTypeFileHeader + "CC", contigFilename + "_Double.fasta", contigFilename + "_Double.fasta"
if True:
alignerRobot.useMummerAlign(mummerLink, folderName, header,
referenceFile, queryFile)
lenDicCC = IORobot.obtainLength(folderName,
contigFilename + "_Double.fasta")
dataListCC = alignerRobot.extractMumData(folderName, header + "Out")
dataListCC = abunHouseKeeper.filterData(dataListCC, lenDicCC)
header, referenceFile, queryFile = optTypeFileHeader + "RR", readsetFilename + "_Double.fasta", readsetFilename + "_Double.fasta"
lenDicRR = IORobot.obtainLength(folderName,
readsetFilename + "_Double.fasta")
if not abunHouseKeeper.abunGlobalRRDisable:
if True:
alignerSubRoutine(folderName, referenceFile, queryFile, mummerLink,
header)
dataListRR = alignerRobot.extractMumData(folderName, header + "Out")
dataListRR = abunHouseKeeper.filterData(dataListRR, lenDicRR)
for eachitem in dataListRR:
if eachitem[-2] == "Read164_p" and eachitem[-1] == "Read159_p":
print "debug", eachitem
if eachitem[-1] == "Read164_p" and eachitem[-2] == "Read159_p":
print "debug", eachitem
dataListRR = abunHouseKeeper.filterDataIdentical(dataListRR, lenDicRR)
else:
dataListRR = []
header, referenceFile, queryFile = optTypeFileHeader + "CR", contigFilename + "_Double.fasta", readsetFilename + "_Double.fasta"
if True:
alignerSubRoutine(folderName, referenceFile, queryFile, mummerLink,
header)
#alignerRobot.useMummerAlign(mummerLink, folderName, header, referenceFile, queryFile)
lenDicCR = dict(lenDicCC.items() + lenDicRR.items())
dataListCR = alignerRobot.extractMumData(folderName, header + "Out")
dataListCR = abunHouseKeeper.filterData(dataListCR, lenDicCR)
numberOfNodes = len(lenDicCR)
G = graphLib.seqGraph(numberOfNodes)
N1, N2 = len(lenDicCC), len(lenDicRR)
print "N1, N2, numberOfNodes: ", N1, N2, numberOfNodes
'''
e.g. of dataListCC[0], dataListRR[0], dataListCR[0]
[1, 520, 2913194, 2913716, 520, 523, 99.05, 'Contig0_d', 'Contig2_d']
[1, 1383, 1253, 2603, 1383, 1351, 82.39, 'Read0_d', 'Read1705_p']
[1, 718, 4334, 5074, 718, 741, 91.91, 'Contig0_d', 'Read1018_d']
'''
# print dataListCC[0]
# print dataListRR[0]
# print dataListCR[0]
# for eachitem in dataListCC:
# print eachitem
addDataToList(dataListCC, G, 0, 0, 'C', 'C')
# for eachitem in dataListRR[0:10]:
# print eachitem , lenDicRR[eachitem[-2]], lenDicRR[eachitem[-1]]
addDataToList(dataListRR, G, N1, N1, 'R', 'R')
addDataToList(dataListCR, G, 0, N1, 'C', 'R')
# G.reportEdge()
G.saveToFile(folderName, graphName)
checkGraphLength(G, N1, lenDicRR)
# print len(G.graphNodesList[0].listOfPrevNodes), len(G.graphNodesList[0].listOfNextNodes)
print "len(G.graphNodesList)", len(G.graphNodesList)
