def merge(genome, trackName, allowOverlaps):
path = createDirPath(trackName, genome, allowOverlaps=allowOverlaps)
collector = PreProcMetaDataCollector(genome, trackName)
chrList = collector.getPreProcessedChrs(allowOverlaps)
if not collector.getTrackFormat().reprIsDense():
chrList = sorted(chrList)
existingChrList = [
chr
for chr in ChrMemmapFolderMerger._existingChrIter(path, chrList)
]
if len(existingChrList) == 0:
raise EmptyGESourceError(
'No data lines has been read from source file (probably because it is empty).'
)
firstChrTrackData = TrackSource().getTrackData(trackName,
genome,
existingChrList[0],
allowOverlaps,
forceChrFolders=True)
arrayList = firstChrTrackData.keys()
for arrayName in arrayList:
mergedArray = firstChrTrackData[arrayName][:]
elementDim, dtypeDim = parseMemmapFileFn(
firstChrTrackData[arrayName].filename)[1:3]
del firstChrTrackData[arrayName]
for chr in existingChrList[1:]:
chrTrackData = TrackSource().getTrackData(trackName,
genome,
chr,
allowOverlaps,
forceChrFolders=True)
mergedArray = ChrMemmapFolderMerger.mergeArrays(
mergedArray, np.array(chrTrackData[arrayName][:]))
elementDimNew, dtypeDimNew = parseMemmapFileFn(
chrTrackData[arrayName].filename)[1:3]
elementDim = max(elementDim, elementDimNew)
dtypeDim = max(dtypeDim, dtypeDimNew)
del chrTrackData[arrayName]
mergedFn = createMemmapFileFn(path, arrayName, elementDim,
dtypeDim, str(mergedArray.dtype))
f = np.memmap(mergedFn,
dtype=mergedArray.dtype,
mode='w+',
shape=mergedArray.shape)
f[:] = mergedArray
f.flush()
del f
del mergedArray
def getTrackData(self,
trackName,
genome,
chr,
allowOverlaps,
forceChrFolders=False):
trackData = TrackData()
brShelve = BoundingRegionShelve(genome, trackName, allowOverlaps)
if not forceChrFolders and brShelve.fileExists():
chr = None
dir = createDirPath(trackName, genome, chr, allowOverlaps)
for fn in os.listdir(dir):
fullFn = dir + os.sep + fn
if fn[0] == '.' or os.path.isdir(fullFn):
continue
if isBoundingRegionFileName(fn):
if fullFn not in self._fileDict:
self._fileDict[fullFn] = brShelve
trackData.boundingRegionShelve = self._fileDict[fullFn]
continue
prefix, elementDim, dtypeDim, dtype = parseMemmapFileFn(fn)
assert prefix not in trackData
trackData[prefix] = self._getFile(chr, dir, fullFn, elementDim,
dtype, dtypeDim)
return trackData
def getTrackData(self, trackName, genome, chr, allowOverlaps, forceChrFolders=False):
trackData = TrackData()
brShelve = BoundingRegionShelve(genome, trackName, allowOverlaps)
if not forceChrFolders and brShelve.fileExists():
chr = None
dir = createDirPath(trackName, genome, chr, allowOverlaps)
for fn in os.listdir(dir):
fullFn = dir + os.sep + fn
if fn[0] == '.' or os.path.isdir(fullFn):
continue
if isBoundingRegionFileName(fn):
if fullFn not in self._fileDict:
self._fileDict[fullFn] = brShelve
trackData.boundingRegionShelve = self._fileDict[fullFn]
continue
prefix, elementDim, dtypeDim, dtype = parseMemmapFileFn(fn)
assert prefix not in trackData
trackData[prefix] = self._getFile(chr, dir, fullFn, elementDim, dtype, dtypeDim)
return trackData
def merge(genome, trackName, allowOverlaps):
path = createDirPath(trackName, genome, allowOverlaps=allowOverlaps)
collector = PreProcMetaDataCollector(genome, trackName)
chrList = collector.getPreProcessedChrs(allowOverlaps)
if not collector.getTrackFormat().reprIsDense():
chrList = sorted(chrList)
existingChrList = [chr for chr in ChrMemmapFolderMerger._existingChrIter(path, chrList)]
if len(existingChrList) == 0:
raise EmptyGESourceError('No data lines has been read from source file (probably because it is empty).')
firstChrTrackData = TrackSource().getTrackData(trackName, genome, existingChrList[0], allowOverlaps, forceChrFolders=True)
arrayList = firstChrTrackData.keys()
for arrayName in arrayList:
mergedArray = firstChrTrackData[arrayName][:]
elementDim, dtypeDim = parseMemmapFileFn(firstChrTrackData[arrayName].filename)[1:3]
del firstChrTrackData[arrayName]
for chr in existingChrList[1:]:
chrTrackData = TrackSource().getTrackData(trackName, genome, chr, allowOverlaps, forceChrFolders=True)
mergedArray = ChrMemmapFolderMerger.mergeArrays(mergedArray, np.array(chrTrackData[arrayName][:]))
elementDimNew, dtypeDimNew = parseMemmapFileFn(chrTrackData[arrayName].filename)[1:3]
elementDim = max(elementDim, elementDimNew)
dtypeDim = max(dtypeDim, dtypeDimNew)
del chrTrackData[arrayName]
mergedFn = createMemmapFileFn(path, arrayName, elementDim, dtypeDim, str(mergedArray.dtype))
f = np.memmap(mergedFn, dtype=mergedArray.dtype, mode='w+', shape=mergedArray.shape)
f[:] = mergedArray
f.flush()
del f
del mergedArray
