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 OrthologuesFromTrees(groupsDir, workingDir, nHighParallel, speciesTree_fn = None, pickleDir=None):
"""
groupsDir - directory with orthogroups file in
userSpeciesTree_fn - None if not supplied otherwise rooted tree using user species names (not orthofinder IDs)
workingDir - orthologues 'WorkingDirectory'
qUserSpTree - is the speciesTree_fn user-supplied
Just infer orthologues from trees, don't do any of the preceeding steps.
"""
# Check species tree
qUserSpTree = (speciesTree_fn != None)
if qUserSpTree:
if not os.path.exists(speciesTree_fn):
print("\nERROR: %s does not exist\n" % speciesTree_fn)
util.Fail()
else:
possibilities = ["SpeciesTree_ids_0_rooted.txt", "SpeciesTree_ids_1_rooted.txt", "SpeciesTree_user_ids.txt"] # etc (only need to determine if unique)
nTrees = 0
for p in possibilities:
fn = workingDir + "Trees_ids/" + p
if os.path.exists(fn):
nTrees += 1
speciesTree_fn = fn
if nTrees == 0:
print("\nERROR: There is a problem with the specified directory. The rooted species tree %s or %s is not present." % (possibilities[0], possibilities[2]))
print("Please rectify the problem or alternatively use the -s option to specify the species tree to use.\n")
util.Fail()
if nTrees > 1:
print("\nERROR: There is more than one rooted species tree in the specified directory structure. Please use the -s option to specify which species tree should be used\n")
util.Fail()
def TreePatIDs(iog):
return workingDir + ("Trees_ids/OG%07d_tree_id.txt" % iog)
reconTreesRenamedDir = workingDir + "Recon_Gene_Trees/"
resultsDir_new = workingDir + "../Orthologues" # for the Orthologues_Species/ directories
# if os.path.exists(resultsDir_new):
resultsDir_new = util.CreateNewWorkingDirectory(resultsDir_new + "_")
# else:
# resultsDir_new += os.sep
# os.mkdir(resultsDir_new)
orthofinderWorkingDir, orthofinderResultsDir, clustersFilename_pairs = util.GetOGsFile(groupsDir)
speciesToUse, nSpAll = util.GetSpeciesToUse(orthofinderWorkingDir + "SpeciesIDs.txt")
ogSet = OrthoGroupsSet(orthofinderWorkingDir, speciesToUse, nSpAll, clustersFilename_pairs, idExtractor = util.FirstWordExtractor)
if qUserSpTree:
speciesToUseNames = ogSet.SpeciesDict().values()
CheckUserSpeciesTree(speciesTree_fn, speciesToUseNames)
speciesTree_fn = ConvertUserSpeciesTree(workingDir + "Trees_ids/", speciesTree_fn, ogSet.SpeciesDict())
util.PrintUnderline("Running Orthologue Prediction", True)
util.PrintUnderline("Reconciling gene and species trees")
ReconciliationAndOrthologues(TreePatIDs, ogSet, speciesTree_fn, workingDir, resultsDir_new, reconTreesRenamedDir, nHighParallel, pickleDir=pickleDir)
util.PrintUnderline("Writing results files")
CleanWorkingDir(workingDir)
return "Species-by-species orthologues directory:\n %s\n" % resultsDir_new
def OrthologuesFromTrees(recon_method, nHighParallel, userSpeciesTree_fn, qAddSpeciesToIDs):
"""
userSpeciesTree_fn - None if not supplied otherwise rooted tree using user species names (not orthofinder IDs)
qUserSpTree - is the speciesTree_fn user-supplied
Just infer orthologues from trees, don't do any of the preceeding steps.
"""
speciesToUse, nSpAll, _ = util.GetSpeciesToUse(files.FileHandler.GetSpeciesIDsFN())
ogSet = OrthoGroupsSet(files.FileHandler.GetWorkingDirectory1_Read(), speciesToUse, nSpAll, qAddSpeciesToIDs, idExtractor = util.FirstWordExtractor)
if userSpeciesTree_fn != None:
speciesDict = files.FileHandler.GetSpeciesDict()
speciesToUseNames = [speciesDict[str(iSp)] for iSp in ogSet.speciesToUse]
CheckUserSpeciesTree(userSpeciesTree_fn, speciesToUseNames)
speciesTreeFN_ids = files.FileHandler.GetSpeciesTreeIDsRootedFN()
ConvertUserSpeciesTree(userSpeciesTree_fn, speciesDict, speciesTreeFN_ids)
util.PrintUnderline("Running Orthologue Prediction", True)
util.PrintUnderline("Reconciling gene and species trees")
ReconciliationAndOrthologues(recon_method, ogSet, nHighParallel)
util.PrintUnderline("Writing results files")
util.PrintTime("Writing results files")
files.FileHandler.CleanWorkingDir2()
return "Species-by-species orthologues directory:\n %s\n" % files.FileHandler.GetOrthologuesDirectory()
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
print("Missing option for command line argument -t")
util.Fail()
arg = args.pop(0)
try:
nProcesses = int(arg)
except:
print("Incorrect argument for number of threads: %s" % arg)
util.Fail()
else:
userDir = arg
# Check arguments
print("0. Getting Orthologues")
print("----------------------")
if nProcesses == None:
print(
"""\nNumber of parallel processes has not been specified, will use the default value.
Number of parallel processes can be specified using the -t option.""")
nProcesses = util.nThreadsDefault
print("Using %d threads for alignments and trees" % nProcesses)
orthofinderWorkingDir, orthofinderResultsDir, clustersFilename_pairs = util.GetOGsFile(
userDir)
speciesToUse, nSpAll = util.GetSpeciesToUse(orthofinderWorkingDir +
"SpeciesIDs.txt")
resultsString = GetOrthologues(orthofinderWorkingDir,
orthofinderResultsDir, speciesToUse, nSpAll,
clustersFilename_pairs, nProcesses)
print(resultsString)
util.PrintCitation()