annotate SMART/Java/Python/runRandomJobs.py @ 70:6b009f1530a8 draft

Deleted tmp file.

import unittest
import os
import time
from optparse import OptionParser
from SMART.Java.Python.ncList.test.MockFindOverlaps_randomExample import MockFindOverlaps_randomExample
from SMART.Java.Python.FindOverlapsOptim import FindOverlapsOptim

if __name__ == '__main__':
    description = "runRandomJobs: create random ref/query files (with size given), and run the jobs on cluster with help of runJobs.sh"

    parser = OptionParser(description = description)
    parser.add_option("-i", "--inputRef", dest="inputRefGff3FileName", action="store", type="string", help="Reference input file [compulsory] [format: file in gff3 format]")
    parser.add_option("-j", "--inputQuery", dest="inputQueryGff3FileName", action="store", type="string", help="Query input file [compulsory] [format: file in gff3 format]")
    parser.add_option("-m", "--inputRefSize", dest="numberOfRefReads", action="store", type="int", help="The number of Reference")
    parser.add_option("-n", "--inputQuerySize", dest="numberOfQReads", action="store", type="int", help="The number of Query")
    parser.add_option("-o", "--output", dest="outputGff3FileName", action="store", type="string", help="output file [compulsory] [format: output file in gff3 format]")
    (options, args) = parser.parse_args()
    
    outputDataName = 'timeResult.dat' 
    fTime = open(outputDataName, 'w')  
    fTime.write('NbRef\tNbQuery\tNbOverlap\ttime\n')   
    chromSize = 100000
    print 'ref size = %d,  query size = %d' %(options.numberOfRefReads, options.numberOfQReads)
    iMFOR_ref = MockFindOverlaps_randomExample(options.inputRefGff3FileName, 'ref', options.numberOfRefReads, chromSize)
    iMFOR_ref.write()
    cmd_ref = 'sort -f -n -k4 -k5.4rn -o %s %s' % (options.inputRefGff3FileName, options.inputRefGff3FileName)
    os.system(cmd_ref)
    iMFOR_query = MockFindOverlaps_randomExample(options.inputQueryGff3FileName,'q', options.numberOfQReads, chromSize)
    iMFOR_query.write()
    cmd_query = 'sort -f -n -k4 -k5.4rn -o %s %s' % (options.inputQueryGff3FileName, options.inputQueryGff3FileName)
    os.system(cmd_query)
    iFOO = FindOverlaps_optim(options.inputRefGff3FileName, options.inputQueryGff3FileName)
    iFOO.setOutputGff3FileName(options.outputGff3FileName)
    
    startTime_optim = time.time()
    iFOO.run()
    iFOO.close()  
    nbOverlap = iFOO.getNbOverlap() 
    endTime_optim = time.time()    
    cmd = 'sort -f -n -k4 -k5.4rn -k9.5 -t ";" -o %s %s' % (options.outputGff3FileName, options.outputGff3FileName)
    os.system(cmd)
    totalTime_optim = endTime_optim - startTime_optim
    print 'we take %s second.' % (totalTime_optim)
    fTime.write('%d\t%d\t%d\t%.2f\n'%(options.numberOfRefReads, options.numberOfQReads, nbOverlap, totalTime_optim))
    iFOO.deletIntermediateFiles()
    fTime.close()