def checkUndirectedEdges(genome, trackName, allowOverlaps): collector = PreProcMetaDataCollector(genome, trackName) if not (collector.getTrackFormat().isLinked() and collector.hasUndirectedEdges()): return complementEdgeWeightDict = {} for chr in collector.getPreProcessedChrs(allowOverlaps): trackSource = TrackSource() trackData = trackSource.getTrackData(trackName, genome, chr, allowOverlaps) ids = trackData['id'] edges = trackData['edges'] weights = trackData.get('weights') for i, id in enumerate(ids): edgesAttr = edges[i][edges[i] != ''] weightsAttr = weights[i][edges[i] != ''] if weights is not None else None PreProcessUtils._adjustComplementaryEdgeWeightDict(complementEdgeWeightDict, id, edgesAttr, weightsAttr) if len(complementEdgeWeightDict) != 0: unmatchedPairs = [] for toId in complementEdgeWeightDict: for fromId in complementEdgeWeightDict[toId]: unmatchedPairs.append((fromId, toId, complementEdgeWeightDict[toId][fromId])) raise InvalidFormatError("Error: All edges are not undirected. The following edges specifications " +\ "are not matched by an opposite edge with equal weight:" + os.linesep +\ os.linesep.join(["from '%s' to '%s'" % (fromId, toId) + \ (" with weight '%s'" % weight if weight != '' else '') \ for fromId, toId, weight in unmatchedPairs]))
def __init__(self, trackName): self.trackName = trackName self._trackSource = TrackSource() self._trackViewLoader = TrackViewLoader() self._trackFormatReq = NeutralTrackFormatReq() self.formatConverters = None self._trackId = None
def checkIfEdgeIdsExist(genome, trackName, allowOverlaps): collector = PreProcMetaDataCollector(genome, trackName) if not collector.getTrackFormat().isLinked(): return uniqueIds = numpy.array([], dtype='S') uniqueEdgeIds = numpy.array([], dtype='S') for chr in collector.getPreProcessedChrs(allowOverlaps): trackSource = TrackSource() trackData = trackSource.getTrackData(trackName, genome, chr, allowOverlaps) uniqueIds = numpy.unique(numpy.concatenate((uniqueIds, trackData['id'][:]))) uniqueEdgeIds = numpy.unique(numpy.concatenate((uniqueEdgeIds, trackData['edges'][:].flatten()))) uniqueIds = uniqueIds[uniqueIds != ''] uniqueEdgeIds = uniqueEdgeIds[uniqueEdgeIds != ''] unmatchedIds = set(uniqueEdgeIds) - set(uniqueIds) if len(unmatchedIds) > 0: raise InvalidFormatError("Error: the following ids specified in the 'edges' column do not exist in the dataset: " + ', '.join(sorted(unmatchedIds)))
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
class Track(object): IS_MEMOIZABLE = True def __new__(cls, trackName): if trackName == [] or trackName is None: return None else: if ExternalTrackManager.isVirtualTrack(trackName): return VirtualMinimalTrack.__new__(VirtualMinimalTrack) else: return object.__new__(cls) def __init__(self, trackName): self.trackName = trackName self._trackSource = TrackSource() self._trackViewLoader = TrackViewLoader() self._trackFormatReq = NeutralTrackFormatReq() self.formatConverters = None self._trackId = None def _getRawTrackView(self, region, borderHandling, allowOverlaps): trackData = self._trackSource.getTrackData(self.trackName, region.genome, region.chr, allowOverlaps) return self._trackViewLoader.loadTrackView(trackData, region, borderHandling, allowOverlaps, self.trackName) def getTrackView(self, region): allowOverlaps = self._trackFormatReq.allowOverlaps() borderHandling = self._trackFormatReq.borderHandling() assert allowOverlaps is not None assert borderHandling is not None origTrackView = self._getRawTrackView(region, borderHandling, allowOverlaps) if self.formatConverters is None: self.formatConverters = getFormatConverters(origTrackView.trackFormat, self._trackFormatReq) if self.formatConverters == []: raise IncompatibleTracksError( prettyPrintTrackName(self.trackName) + " with format: " + str(origTrackView.trackFormat) + ("(" + origTrackView.trackFormat._val + ")" if origTrackView.trackFormat._val else "") + " does not satisfy " + str(self._trackFormatReq) ) if not self.formatConverters[0].canHandle(origTrackView.trackFormat, self._trackFormatReq): raise IncompatibleTracksError( getClassName(self.formatConverters[0]) + " does not support conversion from " + str(origTrackView.trackFormat) + " to " + str(self._trackFormatReq) ) return self.formatConverters[0].convert(origTrackView) def addFormatReq(self, requestedTrackFormat): prevFormatReq = self._trackFormatReq self._trackFormatReq = TrackFormatReq.merge(self._trackFormatReq, requestedTrackFormat) if self._trackFormatReq is None: raise IncompatibleTracksError( str(prevFormatReq) + " is incompatible with additional " + str(requestedTrackFormat) ) def setFormatConverter(self, converterClassName): assert self.formatConverters is None if converterClassName is not None: self.formatConverters = [getFormatConverterByName(converterClassName)] def getUniqueKey(self, genome): assert self.formatConverters is not None and len(self.formatConverters) == 1, "FC: " + str( self.formatConverters ) assert not None in [self._trackFormatReq.allowOverlaps(), self._trackFormatReq.borderHandling()] if not self._trackId: self._trackId = TrackInfo(genome, self.trackName).id return hash( ( tuple(self.trackName), self._trackId, getClassName(self.formatConverters[0]), self.formatConverters[0].VERSION, self._trackFormatReq.allowOverlaps(), self._trackFormatReq.borderHandling(), ) )