def bedGraphToBigWig(chromSizes, bedGraphPath, bigWigPath, sort=True):
"""
takes a bedgraph file, orders it and converts it to
a bigwig file using pyBigWig.
"""
from tempfile import NamedTemporaryFile
from os import remove, system
# Make a list of tuples for the bigWig header, this MUST be sorted identically to the bedGraph file
sort_cmd = cfg.config.get('external_tools', 'sort')
_file = NamedTemporaryFile(delete=False)
for chrom, size in chromSizes:
_file.write(toBytes("{}\t{}\n".format(chrom, size)))
_file.close()
system("LC_ALL=C {} -k1,1 -k2,2n {} > {}.sorted".format(
sort_cmd, _file.name, _file.name))
cl = []
f = open("{}.sorted".format(_file.name))
for line in f:
chrom, chromLen = line.split()
cl.append((chrom, int(chromLen)))
f.close()
remove(_file.name)
remove("{}.sorted".format(_file.name))
# check if the file is empty
if os.stat(bedGraphPath).st_size < 10:
import sys
sys.stderr.write(
"Error: The generated bedGraphFile was empty. Please adjust\n"
"your deepTools settings and check your input files.\n")
exit(1)
if sort:
# temporary file to store sorted bedgraph file
_file = NamedTemporaryFile(delete=False)
tempfilename1 = _file.name
system("LC_ALL=C {} -k1,1 -k2,2n {} > {}".format(
sort_cmd, bedGraphPath, tempfilename1))
bedGraphPath = tempfilename1
bw = pyBigWig.open(bigWigPath, "w")
assert (bw is not None)
# The lack of maxZooms will change the results a bit, perhaps the defaults are better
bw.addHeader(cl, maxZooms=10)
f = open(bedGraphPath)
for line in f:
interval = line.split()
bw.addEntries([interval[0]], [int(interval[1])],
ends=[int(interval[2])],
values=[float(interval[3])])
f.close()
bw.close()
if sort:
remove(tempfilename1)
def bedGraphToBigWig(chromSizes, bedGraphPath, bigWigPath, sort=True):
"""
takes a bedgraph file, orders it and converts it to
a bigwig file using pyBigWig.
"""
from tempfile import NamedTemporaryFile
from os import remove, system
# Make a list of tuples for the bigWig header, this MUST be sorted identically to the bedGraph file
sort_cmd = cfg.config.get('external_tools', 'sort')
_file = NamedTemporaryFile(delete=False)
for chrom, size in chromSizes:
_file.write(toBytes("{}\t{}\n".format(chrom, size)))
_file.close()
system("LC_ALL=C {} -k1,1 -k2,2n {} > {}.sorted".format(sort_cmd, _file.name, _file.name))
cl = []
f = open("{}.sorted".format(_file.name))
for line in f:
chrom, chromLen = line.split()
cl.append((chrom, int(chromLen)))
f.close()
remove(_file.name)
remove("{}.sorted".format(_file.name))
# check if the file is empty
if os.stat(bedGraphPath).st_size < 10:
import sys
sys.stderr.write(
"Error: The generated bedGraphFile was empty. Please adjust\n"
"your deepTools settings and check your input files.\n")
exit(1)
if sort:
# temporary file to store sorted bedgraph file
_file = NamedTemporaryFile(delete=False)
tempfilename1 = _file.name
system("LC_ALL=C {} -k1,1 -k2,2n {} > {}".format(sort_cmd, bedGraphPath, tempfilename1))
bedGraphPath = tempfilename1
bw = pyBigWig.open(bigWigPath, "w")
assert(bw is not None)
# The lack of maxZooms will change the results a bit, perhaps the defaults are better
bw.addHeader(cl, maxZooms=10)
f = open(bedGraphPath)
for line in f:
interval = line.split()
bw.addEntries([interval[0]], [int(interval[1])], ends=[int(interval[2])], values=[float(interval[3])])
f.close()
bw.close()
if sort:
remove(tempfilename1)
def writeBedGraph_worker(chrom,
start,
end,
tileSize,
defaultFragmentLength,
bamOrBwFileList,
func,
funcArgs,
extendPairedEnds=True,
smoothLength=0,
missingDataAsZero=False,
fixed_step=False):
r"""
Writes a bedgraph having as base a number of bam files.
The given func is called to compute the desired bedgraph value
using the funcArgs
tileSize
"""
if start > end:
raise NameError("start position ({0}) bigger than "
"end position ({1})".format(start, end))
coverage = []
for indexFile, fileFormat in bamOrBwFileList:
if fileFormat == 'bam':
bamHandle = bamHandler.openBam(indexFile)
coverage.append(
getCoverageFromBam(bamHandle, chrom, start, end, tileSize,
defaultFragmentLength, extendPairedEnds,
True))
bamHandle.close()
elif fileFormat == 'bigwig':
bigwigHandle = pyBigWig.open(indexFile)
coverage.append(
getCoverageFromBigwig(bigwigHandle, chrom, start, end,
tileSize, missingDataAsZero))
bigwigHandle.close()
# is /dev/shm available?
# working in this directory speeds the process
try:
_file = tempfile.NamedTemporaryFile(dir="/dev/shm", delete=False)
except OSError:
_file = tempfile.NamedTemporaryFile(delete=False)
previousValue = None
lengthCoverage = len(coverage[0])
for tileIndex in range(lengthCoverage):
tileCoverage = []
for index in range(len(bamOrBwFileList)):
if smoothLength > 0:
vectorStart, vectorEnd = getSmoothRange(
tileIndex, tileSize, smoothLength, lengthCoverage)
tileCoverage.append(
np.mean(coverage[index][vectorStart:vectorEnd]))
else:
try:
tileCoverage.append(coverage[index][tileIndex])
except IndexError:
sys.exit(
"Chromosome {} probably not in one of the bigwig "
"files. Remove this chromosome from the bigwig file "
"to continue".format(chrom))
value = func(tileCoverage, funcArgs)
if fixed_step:
writeStart = start + tileIndex * tileSize
writeEnd = min(writeStart + tileSize, end)
try:
_file.write(
toBytes("%s\t%d\t%d\t%.2f\n" %
(chrom, writeStart, writeEnd, value)))
except TypeError:
_file.write(
toBytes("{}\t{}\t{}\t{}\n".format(chrom, writeStart,
writeEnd, value)))
else:
if previousValue is None:
writeStart = start + tileIndex * tileSize
writeEnd = min(writeStart + tileSize, end)
previousValue = value
elif previousValue == value:
writeEnd = min(writeEnd + tileSize, end)
elif previousValue != value:
if not np.isnan(previousValue):
_file.write(
toBytes("{0}\t{1}\t{2}\t{3:g}\n".format(
chrom, writeStart, writeEnd, previousValue)))
previousValue = value
writeStart = writeEnd
writeEnd = min(writeStart + tileSize, end)
if not fixed_step:
# write remaining value if not a nan
if previousValue and writeStart != end and \
not np.isnan(previousValue):
_file.write(
toBytes("{0}\t{1}\t{2}\t{3:g}\n".format(
chrom, writeStart, end, previousValue)))
tempFileName = _file.name
_file.close()
return chrom, start, end, tempFileName
def writeBedGraph_worker(
chrom, start, end, tileSize, defaultFragmentLength,
bamOrBwFileList, func, funcArgs, extendPairedEnds=True, smoothLength=0,
skipZeroOverZero=False, missingDataAsZero=False, fixed_step=False):
r"""
Writes a bedgraph having as base a number of bam files.
The given func is called to compute the desired bedgraph value
using the funcArgs
tileSize
"""
if start > end:
raise NameError("start position ({0}) bigger than "
"end position ({1})".format(start, end))
coverage = []
for indexFile, fileFormat in bamOrBwFileList:
if fileFormat == 'bam':
bamHandle = bamHandler.openBam(indexFile)
coverage.append(getCoverageFromBam(
bamHandle, chrom, start, end, tileSize,
defaultFragmentLength, extendPairedEnds,
True))
bamHandle.close()
elif fileFormat == 'bigwig':
bigwigHandle = pyBigWig.open(indexFile)
coverage.append(
getCoverageFromBigwig(
bigwigHandle, chrom, start, end,
tileSize, missingDataAsZero))
bigwigHandle.close()
# is /dev/shm available?
# working in this directory speeds the process
try:
_file = tempfile.NamedTemporaryFile(dir="/dev/shm", delete=False)
except OSError:
_file = tempfile.NamedTemporaryFile(delete=False)
previousValue = None
lengthCoverage = len(coverage[0])
for tileIndex in range(lengthCoverage):
tileCoverage = []
for index in range(len(bamOrBwFileList)):
if smoothLength > 0:
vectorStart, vectorEnd = getSmoothRange(
tileIndex, tileSize, smoothLength, lengthCoverage)
tileCoverage.append(
np.mean(coverage[index][vectorStart:vectorEnd]))
else:
try:
tileCoverage.append(coverage[index][tileIndex])
except IndexError:
sys.exit("Chromosome {} probably not in one of the bigwig "
"files. Remove this chromosome from the bigwig file "
"to continue".format(chrom))
if skipZeroOverZero and np.sum(tileCoverage) == 0:
previousValue = None
continue
value = func(tileCoverage, funcArgs)
if fixed_step:
writeStart = start + tileIndex * tileSize
writeEnd = min(writeStart + tileSize, end)
try:
_file.write(toBytes("%s\t%d\t%d\t%.2f\n" % (chrom, writeStart,
writeEnd, value)))
except TypeError:
_file.write(toBytes("{}\t{}\t{}\t{}\n".format(chrom, writeStart,
writeEnd, value)))
else:
if previousValue is None:
writeStart = start + tileIndex * tileSize
writeEnd = min(writeStart + tileSize, end)
previousValue = value
elif previousValue == value:
writeEnd = min(writeEnd + tileSize, end)
elif previousValue != value:
if not np.isnan(previousValue):
_file.write(
toBytes("{0}\t{1}\t{2}\t{3:g}\n".format(chrom, writeStart,
writeEnd, previousValue)))
previousValue = value
writeStart = writeEnd
writeEnd = min(writeStart + tileSize, end)
if not fixed_step:
# write remaining value if not a nan
if previousValue and writeStart != end and \
not np.isnan(previousValue):
_file.write(toBytes("{0}\t{1}\t{2}\t{3:g}\n".format(chrom, writeStart,
end, previousValue)))
tempFileName = _file.name
_file.close()
return chrom, start, end, tempFileName