view TEannot_lite.py @ 0:b126ea31824f draft default tip

1st Uploaded

#!/usr/bin/env python

import os
import sys
import time
import glob
import shutil
import ConfigParser
import re

if not "REPET_PATH" in os.environ.keys():
	print("ERROR: no environment variable REPET_PATH")
	sys.exit(1)

if (not "REPET_DB" in os.environ.keys()) or (not "REPET_HOST" in os.environ.keys()) or (not "REPET_PORT" in os.environ.keys()) or (not "REPET_USER" in os.environ.keys()) or (not "REPET_PW" in os.environ.keys()):
	print "ERROR: there is at least one environment database variable missing : REPET_DB, REPET_PORT, REPET_HOST, REPET_USER or REPET_PW"
	sys.exit(1)

if not "REPET_JOB_MANAGER" in os.environ.keys():
	print "ERROR: no environment variable REPET_JOB_MANAGER"
	sys.exit(1)


if not "%s/bin" % os.environ["REPET_PATH"] in os.environ["PATH"]: 
	os.environ["PATH"] = "%s/bin:%s" % (os.environ["REPET_PATH"], os.environ["PATH"])

sys.path.append(os.environ["REPET_PATH"])
if not "PYTHONPATH" in os.environ.keys():
	os.environ["PYTHONPATH"] = os.environ["REPET_PATH"]
else:
	os.environ["PYTHONPATH"] = "%s:%s" % (os.environ["REPET_PATH"], os.environ["PYTHONPATH"])


from commons.core.LoggerFactory import LoggerFactory
from commons.core.checker.RepetException import RepetException
from commons.core.utils.FileUtils import FileUtils
from commons.core.utils.RepetOptionParser import RepetOptionParser
from commons.core.seq.FastaUtils import * #FastaUtils
from commons.core.sql.DbFactory import DbFactory

LOG_DEPTH = "TEannot.pipeline"

class TEannot_lite(object):

	def __init__(self, configFileName = "", fastaFileName = "", libraryFileName = "", verbosity = 0):
		self._configFileName = configFileName
		self._fastaFileName = os.path.abspath(fastaFileName)
		self._libraryFileName = os.path.abspath(libraryFileName)
		self._projectName = time.strftime("%Y%m%d%H%M%S")
		self._outputGff = ""
		self._classif = ""
		#self._maskedThreshold = 80
		self._statsFile = ""
		self._outputMasked = ""
		if "REPET_TMP_DIR" in os.environ.keys():
        	    self._tmp_dir = os.environ["REPET_TMP_DIR"]
		else :
        	    self._tmp_dir = ""
		self._verbosity = verbosity
		self._log = LoggerFactory.createLogger("%s.%s" % (LOG_DEPTH, self.__class__.__name__), self._verbosity)

	def setAttributesFromCommandLine(self):
		description = "This script is a ligth version of TEannot. It writes configuration file and launches TEannot."
		epilog = "Example: TEannot_lite.py -i fastaFileName -l fastaLibraryFileName \n"
		version = "1.1"
		parser = RepetOptionParser(description = description, epilog = epilog, version = version)
		parser.add_option("-i", "--fasta",   dest = "fastaFileName" , action = "store" , type = "string", help ="Input fasta file name ", default = "")
		parser.add_option("-l", "--lib",   dest = "libraryFileName" , action = "store" , type = "string", help ="Input fasta library file name ", default = "")
		parser.add_option("-c", "--withClassif",   dest = "withClassif" , action = "store" ,  type = "string" , metavar="CLASSIFFILE" , help ="[optional] To add classification informations in GFF3 file, please put classif file from TEdenovo step. ", default = "")
		#parser.add_option("-t", "--maskedThreshold",   dest = "maskedThreshold" , action = "store", type = "int", metavar="80", help ="[optional] [default: 80] To choose the threshold of the identity percent for the masked fasta file. ", default = 80)
		parser.add_option("-s", "--stats",  dest="withStats",   action="store_true",help = " Get statistical file in output.", default = False)
		parser.add_option("-o", "--output",  dest="outputLabel"  , action = "store", type = "string", help = " [optional] Label for GFF3 output file", default = "")
		parser.add_option("-v", "--verbosity",   dest = "verbosity",  action = "store", type = "int", metavar="2",   help = "Verbosity [optional] [default: 2]", default = 2)
		options = parser.parse_args()[0]
		self._setAttributesFromOptions(options)

	def _setAttributesFromOptions(self, options):
		self.setConfigFileName("")
		if options.fastaFileName=="":
			print "ERROR : You have to enter an input fasta file"
			print "Example: TEdenovo_lite.py -i fastaFileName \n"
			print "More option : TEdenovo_lite.py --help "
			exit(1)
		else : 
			self._fastaFileName = os.path.abspath(options.fastaFileName)
		if options.libraryFileName=="":
			print "ERROR : You have to enter an input libary fasta file"
			print "Example: TEannot_lite.py -i fastaFileName -l fastaLibraryFileName \n"
			print "More option : TEannot_lite.py --help "
			exit(1)
		else :
			self._libraryFileName = os.path.abspath(options.libraryFileName)
		if options.outputLabel=="":
			fastaBaseName=os.path.abspath(re.search(r'([^\/\\]*)\.[fa|fasta|fsa|fas]',options.fastaFileName).groups()[0])
			options.outputLabel = fastaBaseName
		self._outputGff = os.path.abspath(options.outputLabel+'-%s.gff3'%self._projectName[:8])

		if options.withClassif!='':
			self._classif = os.path.abspath(options.withClassif)

		self._outputMasked = os.path.abspath(options.outputLabel+'-%s.mask'%self._projectName[:8])
		#if options.maskedThreshold : 
		#	self._maskedThreshold = options.maskedThreshold
		if options.withStats : 
			self._statsFile = os.path.abspath(options.outputLabel+'-%s-TEstats.txt'%self._projectName[:8])
		self._verbosity = options.verbosity

	def setConfigFileName(self, configFileName):
		self._configFileName = configFileName
		if not self._configFileName:
			self._configFileName = "TEannot_Galaxy_config_%s" % self._projectName

	def setAttributesFromConfigFile(self, configFileName):
		config = ConfigParser.ConfigParser()
		config.readfp( open(configFileName) )

	def _writeConfigFile(self):
		if FileUtils.isRessourceExists(self._configFileName):
			self._logAndRaise("Configuration file '%s' already exists. Won't be overwritten.")

		shutil.copy("%s/config/TEannot.cfg" % os.environ.get("REPET_PATH"), self._configFileName)
		self.setAttributesFromConfigFile(self._configFileName)

		os.system("sed -i 's|repet_host: <your_MySQL_host>|repet_host: %s|' %s" % (os.environ["REPET_HOST"], self._configFileName))
		os.system("sed -i 's|repet_user: <your_MySQL_login>|repet_user: %s|' %s" % (os.environ["REPET_USER"], self._configFileName))
		os.system("sed -i 's|repet_pw: <your_MySQL_password>|repet_pw: %s|' %s" % (os.environ["REPET_PW"], self._configFileName))
		os.system("sed -i 's|repet_db: <your_MySQL_db>|repet_db: %s|' %s" % (os.environ["REPET_DB"], self._configFileName))
		os.system("sed -i 's|repet_port: 3306|repet_port: %s|' %s" % (os.environ["REPET_PORT"], self._configFileName))
		os.system("sed -i 's|repet_job_manager: SGE|repet_job_manager: %s|' %s" % (os.environ["REPET_JOB_MANAGER"], self._configFileName))
		os.system("sed -i 's|project_name: <your_project_name>|project_name: %s|' %s" % (self._projectName, self._configFileName))
		os.system("sed -i 's|project_dir: <absolute_path_to_your_project_directory>|project_dir: %s|' %s" % (os.getcwd().replace("/", "\/"), self._configFileName))
		os.system("sed -i 's|do_join: yes|do_join: no|' %s" % ( self._configFileName))
		os.system("sed -i 's|add_SSRs: no|add_SSRs: yes|' %s" % ( self._configFileName)) 
		os.system("sed -i 's|gff3_compulsory_match_part: no|gff3_compulsory_match_part: yes|' %s" % ( self._configFileName))
		os.system("sed -i 's|BLR_sensitivity: 3|BLR_sensitivity: 2|' %s" % ( self._configFileName))
		os.system("sed -i 's|tmpDir:|tmpDir: %s|g' %s" % (self._tmp_dir,self._configFileName))
		if self._classif!="" : 
			os.system("sed -i 's|gff3_with_classif_info: no|gff3_with_classif_info: yes|' %s" % ( self._configFileName)) 
			os.system("sed -i 's|classif_table_name: <name_of_TEs_table>|classif_table_name: %s_consensus_classif|' %s" % ( self._projectName,self._configFileName))

	def _mergeOutputGff(self):
		file_out=open(self._outputGff,'w')
		file_out.write('##gff-version 3\n')
		file_out.close()
		directory="%s_GFF3chr/"%self._projectName
		outGffs = glob.glob("%s*.gff3"%directory)
		for outGff in outGffs : 
			os.system("grep -v '#' %s >> %s"%(outGff,self._outputGff))
		os.system("sed -i 's|%s_REPET_TEs|REPET_TEs|g' %s" % (self._projectName,self._outputGff))

	def _launchTEannot(self):
		print "START time: %s" % time.strftime("%Y-%m-%d %H:%M:%S")
		lCmds = []
		lCmds.append( "TEannot.py -P %s -C %s -S 1 -v %i" % (self._projectName, self._configFileName, self._verbosity) )
		lCmds.append( "TEannot.py -P %s -C %s -S 2 -a BLR -v %i" % (self._projectName, self._configFileName, self._verbosity) )
		lCmds.append( "TEannot.py -P %s -C %s -S 2 -a RM -v %i" % (self._projectName, self._configFileName, self._verbosity) )
		lCmds.append( "TEannot.py -P %s -C %s -S 2 -a CEN -v %i" % (self._projectName, self._configFileName, self._verbosity) )
		lCmds.append( "TEannot.py -P %s -C %s -S 2 -a BLR -r -v %i" % (self._projectName, self._configFileName, self._verbosity) ) #
		lCmds.append( "TEannot.py -P %s -C %s -S 2 -a RM -r -v %i" % (self._projectName, self._configFileName, self._verbosity) )  #
		lCmds.append( "TEannot.py -P %s -C %s -S 2 -a CEN -r -v %i" % (self._projectName, self._configFileName, self._verbosity) ) #
		lCmds.append( "TEannot.py -P %s -C %s -S 4 -s TRF -v %i" % (self._projectName, self._configFileName, self._verbosity) )
		lCmds.append( "TEannot.py -P %s -C %s -S 4 -s RMSSR -v %i" % (self._projectName, self._configFileName, self._verbosity) )
		lCmds.append( "TEannot.py -P %s -C %s -S 4 -s Mreps -v %i" % (self._projectName, self._configFileName, self._verbosity) )
		lCmds.append( "TEannot.py -P %s -C %s -S 5 -v %i" % (self._projectName, self._configFileName, self._verbosity) )
		lCmds.append( "TEannot.py -P %s -C %s -S 3 -c BLR+RM+CEN -v %i" % (self._projectName, self._configFileName, self._verbosity) )
		lCmds.append( "TEannot.py -P %s -C %s -S 7 -v %i" % (self._projectName, self._configFileName, self._verbosity) )
		lCmds.append( "TEannot.py -P %s -C %s -S 8 -v %i -o GFF3" % (self._projectName, self._configFileName, self._verbosity) )

		if self._classif!='':
			self._setClassifTable()
	
		for cmd in lCmds:
			returnValue = os.system(cmd)
		if returnValue != 0:
			print "ERROR: command '%s' returned %i" % (cmd, returnValue)
			self._cleanTables()
			sys.exit(1)
	
		print "END time: %s" % time.strftime("%Y-%m-%d %H:%M:%S")


	def _maskFasta(self):
		pathFile = self._outputMasked+"_tmp.path"
		setFile = self._outputMasked+"_tmp.set"
		lCmds = []
		lCmds.append("srptExportTable.py -i %s_chr_allTEs_nr_noSSR_path -C %s -o %s -v %s" % (self._projectName,self._configFileName,pathFile,self._verbosity))	
		lCmds.append("MaskSeqFromCoord.py -i %s -m %s -f path -X -o %s -v %s" % (self._fastaFileName,pathFile,self._outputMasked,self._verbosity))
		lCmds.append("srptExportTable.py -i %s_chr_allSSRs_set -C %s -o %s -v %s " % (self._projectName,self._configFileName, setFile,self._verbosity))
		lCmds.append("MaskSeqFromCoord.py -i %s -m %s -f set -X -o %s_SSRmask.fa -v %s" % (self._outputMasked, setFile, self._outputMasked, self._verbosity))

		for cmd in lCmds:
			returnValue = os.system(cmd)
			if returnValue != 0:
				print "ERROR: command '%s' returned %i" % (cmd, returnValue)
				self._cleanTables()
				sys.exit(1)

		#os.system("rm -f %s"%pathFile) 

	def _createStatsFile(self):
		fastaFile=open(self._fastaFileName)
		fastaLength=FastaUtils.dbCumLength( fastaFile )
		cmd = "PostAnalyzeTELib.py -a 3 -g {0} -p {1}_chr_allTEs_nr_noSSR_path -s {1}_refTEs_seq".format(fastaLength,self._projectName) 
		os.system(cmd)
		cmd = "mv %s_chr_allTEs_nr_noSSR_path.globalAnnotStatsPerTE.txt %s"%(self._projectName,self._statsFile)
		os.system(cmd)
		
	def _setClassifTable(self):
		iDb = DbFactory.createInstance()
		iDb.createTable("%s_consensus_classif" % self._projectName, "classif", self._classif, True)
		iDb.close()

	def _launchListAndDropTables(self):	
		cmd = "ListAndDropTables.py"
		cmd += " -C %s" % self._configFileName
		cmd += " -d '%s'" % self._projectName
		os.system(cmd)

	def _cleanJobsTable(self):
		db = DbFactory.createInstance( configFileName = self._configFileName )  
		sql_cmd="DELETE FROM jobs WHERE groupid like '%s%%';"%self._projectName
		db.execute( sql_cmd )
		db.close()

	def _cleanTables(self):
		self._launchListAndDropTables()	
		self. _cleanJobsTable()

	def run(self):
		os.mkdir(self._projectName)
		os.chdir(self._projectName)
		self._writeConfigFile()
		os.symlink(self._fastaFileName,"%s/%s.fa" %(os.getcwd(),self._projectName)) #creer repertoire projet
		os.symlink(self._libraryFileName,"%s/%s_refTEs.fa" %(os.getcwd(),self._projectName))
		self._launchTEannot()
		self._mergeOutputGff()
		self._maskFasta()
		if self._statsFile : 
			self._createStatsFile()
		self._cleanTables()

if __name__ == '__main__':
	iTEannot= TEannot_lite()
	iTEannot.setAttributesFromCommandLine()
	iTEannot.run()