def runSam2bam(inFile, outFile, log, index=True, sort=True, delinFile=False, onlyUnique=False, onlyProperPaired=False, filterMQ=0, L=None, threads=1, verbose=False, dry=False):
if(delinFile and files_exist(outFile) and not files_exist(inFile)):
print("Skipping sam2bam for " + outFile, file=log)
else:
if(onlyUnique and filterMQ == 0):
filterMQ = 1;
success = True
cmd = [getBinary("samtools"), "view", "-@", str(threads), "-Sb", "-o", outFile, inFile]
if filterMQ > 0:
cmd+=["-q", str(filterMQ)]
if onlyProperPaired:
cmd+=["-f", "2"]
if not L is None:
cmd+=["-L", L]
run(" ".join(cmd), log, verbose=verbose, dry=dry)
if(sort):
tmp = outFile + "_tmp"
if(not dry):
os.rename(outFile, tmp)
run(" ".join([getBinary("samtools"), "sort", "-@", str(threads), "-o", outFile, tmp]), log, verbose=verbose, dry=dry)
if(success):
removeFile(tmp)
if(success and delinFile):
if(not dry):
removeFile(inFile)
if(index):
pysamIndex(outFile)
def runSam2bam(inFile,
outFile,
log,
index=True,
sort=None,
delinFile=False,
onlyUnique=False,
onlyProperPaired=False,
filterMQ=0,
L=None,
threads=1,
verbose=False,
dry=False):
if delinFile and files_exist(outFile) and not files_exist(inFile):
print("Skipping sam2bam for %s" % outFile, file=log)
else:
if onlyUnique and filterMQ == 0:
filterMQ = 1
success = True
cmd = [
"samtools view", "-@",
str(threads), "-Sb", "-o", outFile, inFile
]
if filterMQ > 0:
cmd += ["-q", str(filterMQ)]
if onlyProperPaired:
cmd += ["-f", "2"]
if L is not None:
cmd += ["-L", L]
run(" ".join(cmd), log, verbose=verbose, dry=dry)
if sort is not None:
tmp = outFile + "_tmp"
if not dry:
os.rename(outFile, tmp)
if sort.lower() == "index":
run(" ".join(
["samtools sort", "-@",
str(threads), "-o", outFile, tmp]),
log,
verbose=verbose,
dry=dry)
elif sort.lower() == "name":
run(" ".join([
"samtools sort -n", "-@",
str(threads), "-o", outFile, tmp
]),
log,
verbose=verbose,
dry=dry)
if success:
removeFile(tmp)
if success and delinFile:
if not dry:
removeFile(inFile)
if index:
pysamIndex(outFile)
def bamSort(outputBAM, log, newHeader, verbose):
tmp = outputBAM + "_tmp"
if(newHeader != None):
pyOutputBAM = pysam.AlignmentFile(outputBAM, "rb")
pyTmp = pysam.AlignmentFile(tmp, "wb", header=newHeader)
for read in pyOutputBAM:
pyTmp.write(read)
pyOutputBAM.close()
pyTmp.close()
else:
os.rename(outputBAM, tmp)
#run(" ".join(["samtools", "sort", "-@", str(threads) , tmp, replaceExtension(outFile, "")]), log, verbose=verbose, dry=dry)
run(" ".join([getBinary("samtools"), "sort", "-o", outputBAM, tmp]), log, verbose=verbose, dry=False)
#pysam.sort(tmp, outputBAM) # @UndefinedVariable
removeFile(tmp)
def bamSort(outputBAM, log, newHeader, verbose):
tmp = outputBAM + "_tmp"
if(newHeader != None):
pyOutputBAM = pysam.AlignmentFile(outputBAM, "rb")
pyTmp = pysam.AlignmentFile(tmp, "wb", header=newHeader)
for read in pyOutputBAM:
pyTmp.write(read)
pyOutputBAM.close()
pyTmp.close()
else:
os.rename(outputBAM, tmp)
#run(" ".join(["samtools", "sort", "-@", str(threads) , tmp, replaceExtension(outFile, "")]), log, verbose=verbose, dry=dry)
run(" ".join([getBinary("samtools"), "sort", "-o", outputBAM, tmp]), log, verbose=verbose, dry=False)
#pysam.sort(tmp, outputBAM) # @UndefinedVariable
removeFile(tmp)
def bamSort(outputBAM, log, newHeader, paired, verbose):
tmp = outputBAM + "_tmp"
if newHeader is not None:
pyOutputBAM = pysam.AlignmentFile(outputBAM, "rb")
pyTmp = pysam.AlignmentFile(tmp, "wb", header=newHeader)
for read in pyOutputBAM:
pyTmp.write(read)
pyOutputBAM.close()
pyTmp.close()
else:
os.rename(outputBAM, tmp)
if not paired:
run("samtools sort %s -o %s" % (tmp, outputBAM),
log,
verbose=verbose,
dry=False)
else:
run("samtools sort -n %s -o %s" % (tmp, outputBAM),
log,
verbose=verbose)
removeFile(tmp)