def phaseVCF(vcfpath, phasevcfpath):
beagle_path = params.GetSoftwarePath()[1]
java_path = params.GetSoftwarePath()[0]
print(" ___ phasing vcf file ___ ")
if (not vcfpath.endswith('.vcf.gz')):
gzipFile(vcfpath)
vcfpath = vcfpath + '.gz'
path, vcffn = os.path.split(vcfpath)
path2, vcffn2 = os.path.split(phasevcfpath)
phasevcffn = sub('.vcf.gz$', '_phased', vcffn)
command = " ".join([
java_path, "-Xmx4g -jar", beagle_path, "gt=" + vcfpath,
"out=" + "/".join([path2, phasevcffn]), "2> beagle.log"
])
runCommand(command)
return phasevcffn
def thinVCF(invcf, outvcf):
vcftools_path = params.GetSoftwarePath()[4]
command = " ".join(
[vcftools_path, "--vcf", invcf, "--thin 50 --out", outvcf, "--recode"])
runCommand(command)
def removeDupSambamba(bamrepairedfinalsortfn, tmpbams_path=''):
bamrepairedfinalmarkedfn = sub('.sorted.bam$', ".marked.bam", bamrepairedfinalsortfn)
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join([sambamba_path, "markdup","--remove-duplicates", "--nthreads", str(4), bamrepairedfinalsortfn, bamrepairedfinalmarkedfn])
runCommand(command)
return bamrepairedfinalmarkedfn
def extractPairedBAMfromROI(inbamfn, bedfn, outbamfn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join([samtools_path, "view -b -f 0x0001 -L", bedfn, inbamfn, ">", outbamfn])
runCommand(command)
def extractPairedReadfromROI(inbamfn, bedfn, outbamfn, flag="either"):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join(
[bedtools_path, "pairtobed -abam", inbamfn, "-b", bedfn, "-type", flag, ">", outbamfn, "2> bedtool.log"])
runCommand(command)
def removeIfEmpty(bamdir, file):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
if file.endswith(".bam"):
command = " ".join([samtools_path, "view", "/".join([bamdir, file]), "| less | head -1 | wc -l"])
nline = subprocess.check_output(command, shell=True)
if os.path.isfile("/".join([bamdir, file])) and (int(nline) == 0):
os.remove("/".join([bamdir, file]))
def splitStrands(inbamfn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
read_strand1sortfn = sub('.bam$', '.read_pos.bam', inbamfn)
read_strand2sortfn = sub('.bam$', '.read_neg.bam', inbamfn)
mapped_all = sub('sorted.bam$', 'mapped_all.bam', inbamfn)
command1 = " ".join([samtools_path, "view -u -h -f 33", inbamfn, ">", read_strand1sortfn])
command2 = " ".join([samtools_path, "view -u -h -f 17", inbamfn, ">", read_strand2sortfn])
runCommand(command1)
runCommand(command2)
def find_unpaired_reads(inbamfn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
unpairedfn = sub('.bam$', '.unpairedfn.bam', inbamfn)
command1 = " ".join([samtools_path, "view -u -h -f 0x0004", inbamfn, ">", unpairedfn])
runCommand(command1)
def sortByName(inbamfn, outbamfn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
if (inbamfn is not None):
command = " ".join([sambamba_path, "sort -n", inbamfn, "-o", outbamfn])
print(command)
runCommand(command)
def splitBamByChr(inbamfn, path, chr):
if (chr is not None):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join([samtools_path, "view -bh", inbamfn, str(chr), ">", "/".join([path, str(chr) + ".bam"])])
runCommand(command)
def subsample(bamfn1, bamfn2, samplingrate=0.5):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join([samtools_path, "view -s", samplingrate, "-b", bamfn1, ">", bamfn2])
runCommand(command)
def extractBAMfromROI_All(inbamfn, bedfn, outbamfn):
java_path, beagle_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath(
)
command = " ".join(
[samtool_path, "view -b -L", bedfn, inbamfn, ">", outbamfn])
runCommand(command)
def mergeSortBamFiles(mergedBamfn, finalbamdir):
java_path, beagle_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath(
)
command = ""
os.chdir(finalbamdir)
matches = []
for root, dirnames, filenames in os.walk(finalbamdir):
for filename in fnmatch.filter(filenames, '*.bam'):
path = os.path.join(root, filename)
if os.path.islink(path):
path = os.path.realpath(path)
if (not matches.__contains__(path)):
matches.append(path)
command = " ".join([path, command])
command2 = " ".join([sambamba_path, "merge", mergedBamfn, command])
runCommand(command2)
def mergeSortBamFiles(mergedBamfn, finalbamdir):
java_path, beagle_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = ""
os.chdir(finalbamdir)
matches = []
num_files = 0
for root, dirnames, filenames in os.walk(finalbamdir):
for filename in fnmatch.filter(filenames, '*.bam'):
path = os.path.join(root, filename)
if os.path.islink(path):
path = os.path.realpath(path)
if (not matches.__contains__(path)):
matches.append(path)
command = " ".join([path, command])
num_files = num_files + 1
if (num_files > 1):
command2 = " ".join([sambamba_path, "merge", mergedBamfn, command])
runCommand(command2)
elif (num_files == 1):
outbam = sub('.bam$', '.sort.bam', str(command.strip()))
sortBam(command, outbam, finalbamdir)
os.remove(str(command.strip()))
def getStrands(inbamfn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
outbamfn_forward = sub('.bam$', '_forward.bam', inbamfn)
outbamfn_reverse = sub('.bam$', '_reverse.bam', inbamfn)
command1 = " ".join([samtools_path, "view -F 0x10", inbamfn, ">", outbamfn_forward])
command2 = " ".join([samtools_path, "view -f 0x10", inbamfn, ">", outbamfn_reverse])
runCommand(command1)
runCommand(command2)
def countReads(inbamfn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
cmd = " ".join([samtools_path, "view", inbamfn, "|wc -l"])
out, err = subprocess.Popen(cmd,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
stdin=subprocess.PIPE, shell=True).communicate()
return "".join(out.split())
def sortIndexBam(inbamfn, outbamfn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join([sambamba_path, "sort", inbamfn, "-o", outbamfn])
command2 = " ".join([sambamba_path, "index", outbamfn])
runCommand(command)
runCommand(command2)
def splitPairAndStrands(inbamfn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
read1_strand1sortfn = sub('.bam$', '.read1_pos.bam', inbamfn)
read1_strand2sortfn = sub('.bam$', '.read1_neg.bam', inbamfn)
read2_strand1sortfn = sub('.bam$', '.read2_pos.bam', inbamfn)
read2_strand2sortfn = sub('.bam$', '.read2_neg.bam', inbamfn)
mapped_all = sub('sorted.bam$', 'mapped_all.bam', inbamfn)
command1 = " ".join([samtools_path, "view -u -h -f 0x0061", inbamfn, ">", read1_strand1sortfn])
command2 = " ".join([samtools_path, "view -u -h -f 0x0051", inbamfn, ">", read1_strand2sortfn])
command3 = " ".join([samtools_path, "view -u -h -f 0x0091", inbamfn, ">", read2_strand1sortfn])
command4 = " ".join([samtools_path, "view -u -h -f 0x00A1", inbamfn, ">", read2_strand2sortfn])
runCommand(command1)
runCommand(command2)
runCommand(command3)
runCommand(command4)
def sortBam(inbamfn, outbamfn, tmpbams_path=''):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join([sambamba_path, "sort", inbamfn, "-o", outbamfn, '--tmpdir=', tmpbams_path])
runCommand(command)
def extract_proper_paired_reads(inbamfn, properfn):
# properfn = sub('.bam$', '_proper.bam', inbamfn)
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join([samtools_path, "view -f 0x03 -bSq 30", inbamfn, ">", properfn])
runCommand(command)
os.remove(inbamfn)
def getProperPairs(inbamfn, outbamfn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join([samtools_path, "view -u -h -f 0x0003", inbamfn, ">", outbamfn])
runCommand(command)
def thinVCF(invcf, outvcf):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join([vcftools_path, "--vcf", invcf, "--thin 50 --out", outvcf, "--recode"])
runCommand(command)
def merge_bams(bamfn1, bamfn2, mergefn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join([sambamba_path, "merge", mergefn, bamfn1, bamfn2, "--nthreads", str(4)])
runCommand(command)
def extractAllReadsfromROI(inbamfn, bedfn, outbamfn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join(
[bedtools_path, "intersect -abam", inbamfn, "-b", bedfn, ">", outbamfn, "2> bedtool.log"])
runCommand(command)
def merge_final(mergefn, finalbamdir):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
mergemarkedfn = sub('.bam$', ".marked.bam", mergefn)
os.chdir(finalbamdir)
command1 = " ".join([sambamba_path, "merge", mergefn, "*.bam", "--nthreads", str(4)])
command2 = " ".join([sambamba_path, "markdup","--remove-duplicates", "--nthreads", str(4), mergefn, mergemarkedfn])
runCommand(command1)
print (" ___ removing merged duplicates near breakpoints ___ ")
runCommand(command2)
os.remove(mergefn)
os.remove(mergefn + '.bai')
os.rename(mergemarkedfn, mergefn)
os.rename(mergemarkedfn + '.bai', mergefn + '.bai')
def dedupBam(inbamfn, outbamfn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join([samtool_path, "rmdup", inbamfn, outbamfn])
runCommand(command)
def mergeSortBamFiles(mergedBamfn, finalbamdir):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = ""
os.chdir(finalbamdir)
matches = []
num_files = 0
for root, dirnames, filenames in os.walk(finalbamdir):
for filename in fnmatch.filter(filenames, '*.bam'):
path = os.path.join(root, filename)
if os.path.islink(path):
path = os.path.realpath(path)
if not matches.__contains__(path):
matches.append(path)
command = " ".join([path, command])
num_files = num_files + 1
if num_files > 1:
command2 = " ".join([sambamba_path, "merge", mergedBamfn, command, "--nthreads", str(4)])
runCommand(command2)
elif num_files == 1:
if str(command.strip()).endswith("GAIN.bam"):
path, fname = os.path.split(str(command.strip()))
inbam_original = '/'.join([params.GetSplitBamsPath(), sub('_gain', '', fname.lower())])
command2 = " ".join([sambamba_path, "merge", mergedBamfn, command, inbam_original, "--nthreads", str(4)])
runCommand(command2)
elif str(command.strip()).endswith("LOSS.bam"):
outbam = sub('.bam$', '.sort.bam', str(command.strip()))
sortBam(command, outbam, finalbamdir)
os.remove(str(command.strip()))
def removeDupPicard(bamrepairedfinalsortfn, tmpbams_path=''):
print (" ___ removing repaired duplicates ___ ")
bamrepairedfinalmarkedfn = sub('.re_paired_final.sorted.bam$', ".re_paired_final.marked.bam", bamrepairedfinalsortfn)
markedmetricsfn = sub('.re_paired_final.sorted.bam$', ".marked_metrics.txt", bamrepairedfinalsortfn)
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
command = " ".join([java_path, "-Xmx8g -jar", picard_path, "MarkDuplicates", "I=" + bamrepairedfinalsortfn, "O=" + bamrepairedfinalmarkedfn, "M=" + markedmetricsfn, "REMOVE_DUPLICATES=true", "ASSUME_SORTED=true"])
runCommand(command)
return bamrepairedfinalmarkedfn
def splitPairs(inbamfn):
java_path, beagle_path, picard_path, samtools_path, bedtools_path, vcftools_path, sambamba_path = params.GetSoftwarePath()
pair1fn = sub('.bam$', '_read1.bam', inbamfn)
pair2fn = sub('.bam$', '_read2.bam', inbamfn)
command1 = " ".join([samtools_path, "view -u -h -f 0x0043", inbamfn, ">", pair1fn])
command2 = " ".join([samtools_path, "view -u -h -f 0x0083", inbamfn, ">", pair2fn])
runCommand(command1)
runCommand(command2)
