def GetSpeciesGenesInfo():
speciesLabels, nSpAll, _ = util.GetSpeciesToUse(
files.FileHandler.GetSpeciesIDsFN())
seqsInfo = util.GetSeqsInfo(files.FileHandler.GetSpeciesSeqsDir(),
speciesLabels, nSpAll)
genenumbers = list(np.diff(seqsInfo.seqStartingIndices))
genenumbers.append(seqsInfo.nSeqs - seqsInfo.seqStartingIndices[-1])
return speciesLabels, genenumbers
def __init__(self, orthofinderWorkingDir_list, speciesToUse, nSpAll, qAddSpeciesToIDs, idExtractor = util.FirstWordExtractor):
self.speciesIDsEx = util.FullAccession(files.FileHandler.GetSpeciesIDsFN())
self._Spec_SeqIDs = None
self._extractor = idExtractor
self.seqIDsEx = None
self.ogs_all = None
self.iOgs4 = 0
self.speciesToUse = speciesToUse # list of ints
self.seqsInfo = util.GetSeqsInfo(orthofinderWorkingDir_list, self.speciesToUse, nSpAll)
self.id_to_og = None
self.qAddSpeciesToIDs = qAddSpeciesToIDs
def __init__(self, orthofinderWorkingDir, speciesToUse, nSpAll, clustersFilename_pairs, idExtractor = util.FirstWordExtractor, pickleDir=None):
self.workingDirOF = orthofinderWorkingDir
self.seqIDsFN = orthofinderWorkingDir + "SequenceIDs.txt"
self.speciesIDsFN = orthofinderWorkingDir + "SpeciesIDs.txt"
self.speciesIDsEx = util.FullAccession(self.speciesIDsFN)
self._Spec_SeqIDs = None
self._extractor = idExtractor
self.clustersFN = clustersFilename_pairs
self.seqIDsEx = None
self.ogs_all = None
self.iOgs4 = 0
self.speciesToUse = speciesToUse
self.seqsInfo = util.GetSeqsInfo(orthofinderWorkingDir, self.speciesToUse, nSpAll)
self.fileInfo = util.FileInfo(workingDir = orthofinderWorkingDir, graphFilename="", separatePickleDir=pickleDir)
self.id_to_og = None
def GetSpeciesGenesInfo(ogSet):
speciesLabels, nSpAll = util.GetSpeciesToUse(ogSet.speciesIDsFN)
seqsInfo = util.GetSeqsInfo(ogSet.workingDirOF, speciesLabels, nSpAll)
genenumbers = list(np.diff(seqsInfo.seqStartingIndices))
genenumbers.append(seqsInfo.nSeqs - seqsInfo.seqStartingIndices[-1])
return speciesLabels, genenumbers
