def copySequenceFiles(self, srcDataDir):
try:
print("Copy Fasta Files from %s to %s" %(srcDataDir,self.orgFastaDir))
orgListFile_fh = open(self.orgListFile)
for line in orgListFile_fh:
organismNameID, organismName = line.strip().split()
if not os.path.exists( NGS_Util.createFilePath(self.orgFastaDir, organismName + ".faa") ):
orgFasta = NGS_Util.createFilePath(srcDataDir, organismName + ".faa")
NGS_Util.copyFile(orgFasta, self.orgFastaDir)
print("Copied fasta file for %s" % (organismName))
else:
print("\tDoing nothing (files already copied) for %s" % (organismName))
orgListFile_fh.close()
except Exception:
print traceback.print_exc()
def getIPRScanScore(self, mode):
try:
print "getIPRScanScore"
orgListFile_fh = open(self.orgListFile)
for line in orgListFile_fh:
organismNameID, organismName = line.strip().split()
organism_IPR_final = NGS_Util.createFilePath(self.fungi_InterProScan_result, organismName + ".faa.IPR.final.txt")
if not os.path.exists(organism_IPR_final):
print "getIPRScanScore : " + organismName
if mode == 1:
self.runClusterIPRScan(organismName)
time.sleep(21600) # sleep for 6 hrs
elif mode == 2:
ipr_xml_file = self.concatenate_ClusterIPRScan_results(organismName)
ipr_raw_file = self.xmlIPRScanToRAWOutput(organismName, ipr_xml_file)
organism_ipr2go = self.extract_ipr2go_based_on_xml(organismName, ipr_xml_file)
organism_ipr2ec = self.map_ipr_to_specific_ecs(organismName, organism_ipr2go)
organism_seqid2ec = self.combine_iprscan_raw_result_with_ipr2ec( organismName, organism_ipr2ec, ipr_raw_file)
if os.path.exists(ipr_raw_file) and os.path.exists(organism_seqid2ec):
organism_raw_final = NGS_Util.createFilePath(self.fungi_InterProScan_result, organismName + ".faa.raw.txt")
organism_IPR_final = NGS_Util.createFilePath(self.fungi_InterProScan_result, organismName + ".faa.IPR.final.txt")
NGS_Util.copyFile(ipr_raw_file, organism_raw_final)
NGS_Util.copyFile(organism_seqid2ec, organism_IPR_final)
orgListFile_fh.close()
except Exception:
print traceback.print_exc()
return ""
def getIPRScanScore(self):
try:
orgListFile_fh = open(self.orgListFile)
for line in orgListFile_fh:
if line.startswith("#"):
continue
organismNameID, organismName = line.strip().split()
organism_IPR_final = NGS_Util.createFilePath(self.fungi_InterProScan_result, organismName + ".faa.IPR.final.txt")
# self.create_new_seq_org_list(organismName,organismNameID)
if not os.path.exists(organism_IPR_final):
print "getIPRScanScore : " + organismName
org_ipr_split_dir = self.splitFiles(organismName)
ipr_raw_file = self.rawIPRScan(organismName,org_ipr_split_dir)
ipr_xml_file = self.rawIPRScanToXMlOutput( organismName, ipr_raw_file)
organism_ipr2go = self.extract_ipr2go_based_on_xml(organismName, ipr_xml_file)
organism_ipr2ec = self.map_ipr_to_specific_ecs(organismName, organism_ipr2go)
organism_seqid2ec = self.combine_iprscan_raw_result_with_ipr2ec( organismName, organism_ipr2ec, ipr_raw_file)
if os.path.exists(ipr_raw_file) and os.path.exists(organism_seqid2ec):
organism_raw_final = NGS_Util.createFilePath(self.fungi_InterProScan_result, organismName + ".faa.raw.txt")
organism_IPR_final = NGS_Util.createFilePath(self.fungi_InterProScan_result, organismName + ".faa.IPR.final.txt")
NGS_Util.copyFile(ipr_raw_file, organism_raw_final)
NGS_Util.copyFile(organism_seqid2ec, organism_IPR_final)
orgListFile_fh.close()
except Exception:
print traceback.print_exc()
return ""
def initialize(self, seq_org_list, jointBlastDir, GTGFungiKNNDir, fungi_InterProScan_result, phylogeneticTreeFile, modelTrainingDir):
try:
self.seq_org_list = seq_org_list
self.jointBlastDir = jointBlastDir
self.GTGFungiKNNDir = GTGFungiKNNDir
self.fungi_InterProScan_result = fungi_InterProScan_result
self.phylogeneticTreeFile = phylogeneticTreeFile
self.modelTrainingDir = modelTrainingDir
self.modelTraining_IPR_EC_Dir = NGS_Util.createDirectoryPath(self.modelTrainingDir, "IPR_EC")
self.modelTrainingBlastPVDir = NGS_Util.createDirectoryPath(self.modelTrainingDir, "BlastPValues")
self.modelTraining_EC_Scores_Dir = NGS_Util.createDirectoryPath(self.modelTrainingDir, "ECScores")
self.modelTrainingProbabilityDensityScoreDir = NGS_Util.createDirectoryPath(self.modelTrainingDir, "ProbabilityDensityScore")
self.modelTrainingTreeDir = NGS_Util.createDirectoryPath(self.modelTrainingDir, "Tree")
self.modelTrainingModelDir = NGS_Util.createDirectoryPath(self.modelTrainingDir, "Model")
NGS_Util.createDirectory(self.modelTrainingDir)
NGS_Util.createDirectory(self.modelTraining_IPR_EC_Dir)
NGS_Util.createDirectory(self.modelTrainingBlastPVDir)
NGS_Util.createDirectory(self.modelTraining_EC_Scores_Dir)
NGS_Util.createDirectory(self.modelTrainingProbabilityDensityScoreDir)
NGS_Util.createDirectory(self.modelTrainingTreeDir)
NGS_Util.createDirectory(self.modelTrainingModelDir)
if (os.path.exists(self.phylogeneticTreeFile)):
NGS_Util.copyFile( self.phylogeneticTreeFile,NGS_Util.createFilePath(self.modelTrainingTreeDir,"tree") )
self.phylogeneticTreeFile = NGS_Util.createFilePath(self.modelTrainingTreeDir,"tree")
self.treeCPDS = NGS_Util.createFilePath(self.modelTrainingTreeDir,"tree.cpds")
except Exception:
print traceback.print_exc()