combineRPKMs.py

   1 #
   2 #  combineRPKMS.py
   3 #  ENRAGE
   4 #
   5
   6 print 'version 1.0'
   7 try:
   8     import psyco
   9     psyco.full()
  10 except:
  11     pass
  12
  13 import sys, optparse
  14
  15
  16 def main(argv=None):
  17     if not argv:
  18         argv = sys.argv
  19
  20     usage = "usage: python %prog firstRPKM expandedRPKM finalRPKM combinedOutfile [--withmultifraction]"
  21     parser = optparse.OptionParser(usage=usage)
  22     parser.add_option("--withmultifraction", action="store_true", dest="doFraction")
  23     parser.set_defaults(doFraction=False)
  24     (options, args) = parser.parse_args(argv[1:])
  25
  26     if len(args) < 3:
  27         print usage
  28         sys.exit(1)
  29
  30     firstfile = args[0]
  31     expandedfile = args[1]
  32     finalfile = args[2]
  33     outfile = args[3]
  34
  35     combineRPKMs(firstfile, expandedfile, finalfile, outfile, options.doFraction)
  36
  37
  38 def combineRPKMs(firstfileName, expandedfileName, finalfileName, outfileName, doFraction=False):
  39     firstfile = open(firstfileName)
  40     expandedfile = open(expandedfileName)
  41     finalfile = open(finalfileName)
  42     outfile = open(outfileName, "w")
  43
  44     firstDict = {}
  45     gidDict = {}
  46     expandedDict = {}
  47
  48     for line in firstfile:
  49         fields = line.strip().split()
  50         firstDict[fields[1]] = fields[-1]
  51
  52     firstfile.close()
  53
  54     for line in expandedfile:
  55         fields = line.strip().split()
  56         expandedDict[fields[1]] = fields[-1]
  57         gidDict[fields[1]] = fields[0]
  58
  59     expandedfile.close()
  60
  61     if doFraction:
  62         header = "gid\tRNAkb\tgene\tfirstRPKM\texpandedRPKM\tfinalRPKM\tfractionMulti\n"
  63     else:
  64         header = "gid\tRNAkb\tgene\tfirstRPKM\texpandedRPKM\tfinalRPKM\n"
  65
  66     outfile.write(header)
  67
  68     for line in finalfile:
  69         fields = line.strip().split()
  70         gene = fields[0]
  71         rnakb = fields[1]
  72         finalRPKM = fields[2]
  73         firstRPKM = firstDict.get(gene, "")
  74         outline = "%s\t%s\t%s\t%s\t%s\t%s" % (gidDict[gene], rnakb, gene, firstRPKM, expandedDict[gene], finalRPKM)
  75
  76         if doFraction:
  77             fraction = fields[3]
  78             outline += "\t%s" % fraction
  79
  80         outfile.write(outline + '\n')
  81
  82     finalfile.close()
  83     outfile.close()
  84
  85
  86 if __name__ == "__main__":
  87     main(sys.argv)