def Map(inputBAM, inputReference, outputSAM, log, quantseqMapping, endtoendMapping, threads=1, parameter="--no-progress --slam-seq 2" , outputSuffix="_ngm_slamdunk", trim5p=0, maxPolyA=-1, topn=1, sampleId=None, sampleName="NA", sampleType="NA", sampleTime=0, printOnly=False, verbose=True, force=False):
if(quantseqMapping is True) :
parameter = "--no-progress"
if(trim5p > 0):
parameter = parameter + " -5 " + str(trim5p)
if(maxPolyA > -1):
parameter = parameter + " --max-polya " + str(maxPolyA)
if(endtoendMapping is True):
parameter = parameter + " -e "
else:
parameter = parameter + " -l "
if(sampleId != None):
parameter = parameter + " --rg-id " + str(sampleId)
if(sampleName != ""):
parameter = parameter + " --rg-sm " + sampleName + ":" + sampleType + ":" + str(sampleTime)
if(topn > 1):
parameter = parameter + " -n " + str(topn) + " --strata "
if(checkStep([inputReference, inputBAM], [replaceExtension(outputSAM, ".bam")], force)):
if outputSAM.endswith(".sam"):
# Output SAM
run(getBinary("ngm") + " -r " + inputReference + " -q " + inputBAM + " -t " + str(threads) + " " + parameter + " -o " + outputSAM, log, verbose=verbose, dry=printOnly)
else:
# Output BAM directly
run(getBinary("ngm") + " -b -r " + inputReference + " -q " + inputBAM + " -t " + str(threads) + " " + parameter + " -o " + outputSAM, log, verbose=verbose, dry=printOnly)
else:
print("Skipped mapping for " + inputBAM, file=log)
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 simulateReads(bed12, bed12Fasta, explv, bedReads, faReads, readLength, readCount, seqError):
#output = shell(getBinary("gensimreads.py") + " -l " + str(readLength) + " -e " + explv + " -n " + str(readCount) + " -b " + rNASeqReadSimulatorPath + "demo/input/sampleposbias.txt --stranded " + bed12 + " > " + bedReads)
output = shell(getBinary("gensimreads.py") + " -l " + str(readLength) + " -e " + explv + " -n " + str(readCount) + " --stranded " + bed12 + " 2> /dev/null > " + bedReads)
if len(output.strip()) > 5:
print(output)
output = shell(getBinary("getseqfrombed.py") + " -f -r " + str(seqError) + " -l " + str(readLength) + " " + bedReads + " " + bed12Fasta + " 2> /dev/null > " + faReads)
if len(output.strip()) > 5:
print(output)
def simulateReads(bed12, bed12Fasta, explv, bedReads, faReads, readLength,
readCount, seqError):
#output = shell(getBinary("gensimreads.py") + " -l " + str(readLength) + " -e " + explv + " -n " + str(readCount) + " -b " + rNASeqReadSimulatorPath + "demo/input/sampleposbias.txt --stranded " + bed12 + " > " + bedReads)
output = shell(
getBinary("gensimreads.py") + " -l " + str(readLength) + " -e " +
explv + " -n " + str(readCount) + " --stranded " + bed12 +
" 2> /dev/null > " + bedReads)
if len(output.strip()) > 5:
print(output)
output = shell(
getBinary("getseqfrombed.py") + " -f -r " + str(seqError) + " -l " +
str(readLength) + " " + bedReads + " " + bed12Fasta +
" 2> /dev/null > " + faReads)
if len(output.strip()) > 5:
print(output)
def checkNextGenMapVersion():
ngmHelp = shellerr(getBinary("ngm"), raiseError = False)
matchObj = re.match( r'.*([0-9]+\.[0-9]+\.[0-9]+).*', ngmHelp, re.M|re.I)
if matchObj:
version = matchObj.group(1)
if version != __ngm_version__:
raise RuntimeError('NextGenMap version expected: ' + __ngm_version__ + " but found " + version + ". Please reinstall slamdunk package.")
else:
raise RuntimeError('Could not get NextGenMap version. Please reinstall slamdunk package.')
def SNPs(inputBAM, outputSNP, referenceFile, minVarFreq, minCov, minQual, log, printOnly=False, verbose=True, force=False):
if(checkStep([inputBAM, referenceFile], [outputSNP], force)):
fileSNP = open(outputSNP, 'w')
mpileupCmd = getBinary("samtools") + " mpileup -B -A -f " + referenceFile + " " + inputBAM
if(verbose):
print(mpileupCmd, file=log)
if(not printOnly):
mpileup = subprocess.Popen(mpileupCmd, shell=True, stdout=subprocess.PIPE, stderr=log)
varscanCmd = "java -jar " + getBinary("VarScan.v2.4.1.jar") + " mpileup2snp --strand-filter 0 --output-vcf --min-var-freq " + str(minVarFreq) + " --min-coverage " + str(minCov) + " --variants 1"
if(verbose):
print(varscanCmd, file=log)
if(not printOnly):
varscan = subprocess.Popen(varscanCmd, shell=True, stdin=mpileup.stdout, stdout=fileSNP, stderr=log)
varscan.wait()
fileSNP.close()
else:
print("Skipping SNP calling", file=log)
def SNPs(inputBAM, outputSNP, referenceFile, minVarFreq, minCov, minQual, log, printOnly=False, verbose=True, force=False):
if(checkStep([inputBAM, referenceFile], [outputSNP], force)):
fileSNP = open(outputSNP, 'w')
mpileupCmd = getBinary("samtools") + " mpileup -B -A -f " + referenceFile + " " + inputBAM
if(verbose):
print(mpileupCmd, file=log)
if(not printOnly):
mpileup = subprocess.Popen(mpileupCmd, shell=True, stdout=subprocess.PIPE, stderr=log)
varscanCmd = "java -jar " + getBinary("VarScan.v2.4.1.jar") + " mpileup2snp --strand-filter 0 --output-vcf --min-var-freq " + str(minVarFreq) + " --min-coverage " + str(minCov) + " --variants 1"
if(verbose):
print(varscanCmd, file=log)
if(not printOnly):
varscan = subprocess.Popen(varscanCmd, shell=True, stdin=mpileup.stdout, stdout=fileSNP, stderr=log)
varscan.wait()
fileSNP.close()
else:
print("Skipping SNP calling", file=log)
def prepareUTRs(bed, bed12, bed12Fasta, referenceFasta, readLength, polyALength, explv, snpRate, vcfFile):
# Read utrs from BED file
utrs = parseUtrBedFile(bed)
vcf = open(vcfFile, "w")
print("##fileformat=VCFv4.1", file=vcf)
print("#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO", file=vcf)
bedFile = BedTool(bed)
bedFasta = bedFile.sequence(fi=referenceFasta, s=True, name=True)
bed12FastaFile = open(bed12Fasta, "w")
utrName = None
for line in bedFasta.print_sequence().splitlines():
if(line[0] == ">"):
print(line, file=bed12FastaFile)
utrName = line[1:]
else:
print(simulateUTR(line, utrs[utrName], polyALength, snpRate, vcf), file=bed12FastaFile)
bed12FastaFile.close()
vcf.close()
bed12File = open(bed12, "w")
totalLength = 0
minFragmentLength = 150
maxFragmentLength = 450
for utr in BedIterator(bed):
fragmentLength = random.randrange(minFragmentLength, maxFragmentLength, 1) #+ readLength
fragmentLength = min(fragmentLength, utr.getLength())
start = max(0, utr.getLength() - fragmentLength)
end = utr.getLength() #- readLength
totalLength += (end - start)
# min(utr.getLength() + readLength / 4, fragmentLength + readLength / 4)
print(utr.name, start, end, utr.name, utr.score, "+", start, end, "255,0,0", "1", (end - start), 0, sep="\t", file=bed12File)
bed12File.close()
output = shell(getBinary("genexplvprofile.py") + " --geometric 1 " + bed12 + " 2> /dev/null > " + explv)
if len(output.strip()) > 5:
print(output)
return totalLength
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 prepareUTRs(bed, bed12, bed12Fasta, referenceFasta, readLength,
polyALength, explv, snpRate, vcfFile):
# Read utrs from BED file
utrs = parseUtrBedFile(bed)
vcf = open(vcfFile, "w")
print("##fileformat=VCFv4.1", file=vcf)
print("#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO", file=vcf)
bedFile = BedTool(bed)
bedFasta = bedFile.sequence(fi=referenceFasta, s=True, name=True)
bed12FastaFile = open(bed12Fasta, "w")
utrName = None
for line in bedFasta.print_sequence().splitlines():
if (line[0] == ">"):
print(line, file=bed12FastaFile)
utrName = line[1:]
else:
print(simulateUTR(line, utrs[utrName], polyALength, snpRate, vcf),
file=bed12FastaFile)
bed12FastaFile.close()
vcf.close()
bed12File = open(bed12, "w")
totalLength = 0
minFragmentLength = 150
maxFragmentLength = 450
for utr in BedIterator(bed):
fragmentLength = random.randrange(minFragmentLength, maxFragmentLength,
1) #+ readLength
fragmentLength = min(fragmentLength, utr.getLength())
start = max(0, utr.getLength() - fragmentLength)
end = utr.getLength() #- readLength
totalLength += (end - start)
# min(utr.getLength() + readLength / 4, fragmentLength + readLength / 4)
print(utr.name,
start,
end,
utr.name,
utr.score,
"+",
start,
end,
"255,0,0",
"1", (end - start),
0,
sep="\t",
file=bed12File)
bed12File.close()
output = shell(
getBinary("genexplvprofile.py") + " --geometric 1 " + bed12 +
" 2> /dev/null > " + explv)
if len(output.strip()) > 5:
print(output)
return totalLength
