regiontobed.py

   1 """
   2     usage: python regiontobed label regionfile outbedfile [--color r,g,b] [--score field] [--narrowPeak] [--broadPeak] [--itemRgb] [--nolabel]
   3            where color is in comma-delimited RGB without space
   4            and field is a column with a score (first column is 0, second is 1,...)
   5            t-narrowPeak assumes that findall.py was run with -listPeak
   6            t-broadPeak assumes that findall.py was *NOT* run with -listPeak
   7 """
   8
   9 try:
  10     import psyco
  11     psyco.full()
  12 except:
  13     pass
  14
  15 import sys
  16 import math
  17 import optparse
  18 from commoncode import getConfigParser, getConfigOption, getConfigBoolOption
  19
  20 print "regiontobed: version 3.2"
  21
  22
  23 def usage():
  24     print __doc__
  25
  26
  27 def main(argv=None):
  28     if not argv:
  29         argv = sys.argv
  30
  31     usage = __doc__
  32
  33     parser = getParser(usage)
  34     (options, args) = parser.parse_args(argv[1:])
  35
  36     if len(args) < 3:
  37         usage()
  38         sys.exit(2)
  39
  40     factorlabel = args[0]
  41     regionfile = args[1]
  42     outfile = args[2]
  43
  44     regiontobed(factorlabel, regionfile, outfile, options.color,
  45                 options.scoreField, options.doNarrow, options.doBroad,
  46                 options.itemRGB, options.noLabel)
  47
  48
  49 def getParser(usage):
  50     parser = optparse.OptionParser(usage=usage)
  51     parser.add_option("--color", dest="color")
  52     parser.add_option("--score", type="int", dest="scoreField")
  53     parser.add_option("--narrowPeak", action="store_true", dest="doNarrow")
  54     parser.add_option("--broadPeak", action="store_true", dest="doBroad")
  55     parser.add_option("--itemRgb", action="store_true", dest="itemRGB")
  56     parser.add_option("--nolabel", action="store_true", dest="noLabel")
  57
  58     configParser = getConfigParser()
  59     section = "regiontobed"
  60     color = getConfigOption(configParser, section, "color", "0,0,0")
  61     scoreField = getConfigOption(configParser, section, "scoreField", None)
  62     doNarrow = getConfigBoolOption(configParser, section, "doNarrow", False)
  63     doBroad = getConfigBoolOption(configParser, section, "doBroad", False)
  64     itemRGB = getConfigBoolOption(configParser, section, "itemRGB", False)
  65     noLabel = getConfigBoolOption(configParser, section, "noLabel", False)
  66
  67     parser.set_defaults(color=color, scoreField=scoreField, doNarrow=doNarrow,
  68                         doBroad=doBroad, itemRGB=itemRGB, noLabel=noLabel)
  69
  70     return parser
  71
  72
  73 def regiontobed(factorlabel, regionFileName, outFileName, color="0,0,0",
  74                 scoreField=None, doNarrow=False, doBroad=False, itemRGB=False,
  75                 noLabel=False):
  76
  77     regionfile = open(regionFileName)
  78     outfile = open(outFileName, "w")
  79
  80     if itemRGB:
  81         print "assigning each item its color"
  82
  83     if noLabel:
  84         if itemRGB:
  85             outfile.write('track name=%s visibility=4 itemRgb="on"\n' % factorlabel)
  86         else:
  87             outfile.write("track name=%s visibility=4 color=%s\n" % (factorlabel, color))
  88
  89     for line in regionfile:
  90         if line[0] == "#":
  91             continue
  92
  93         fields = line.strip().split()
  94         if doNarrow:
  95             signalVal = float(fields[4])
  96             pval = float(fields[-1])
  97             if pval == 0.:
  98                 pValue = 350
  99             else:
 100                 pValue = -1. * math.log(pval, 10)
 101
 102             peakPos = int(fields[9]) - int(fields[2])
 103             outfile.write("%s\t%s\t%s\t%s\t%d\t.\t%.4f\t%.4f\t-1\t%d" % (fields[1], fields[2], fields[3], fields[0], 0, signalVal, pValue, peakPos))
 104         elif doBroad:
 105             signalVal = float(fields[4])
 106             pval = float(fields[-1])
 107             if pval == 0.:
 108                 pValue = 350
 109             else:
 110                 pValue = -1. * math.log(pval, 10)
 111
 112             outfile.write("%s\t%s\t%s\t%s\t%d\t.\t%.4f\t%.4f\t-1" % (fields[1], fields[2], fields[3], fields[0], 0, signalVal, pValue))
 113         elif scoreField is not None:
 114             score = int(float(fields[scoreField]))
 115             if score > 1000:
 116                 score = 1000
 117
 118             outfile.write("%s\t%s\t%s\t%s\t%s" % (fields[1], fields[2], fields[3], fields[0], score))
 119             if itemRGB:
 120                 outfile.write("\t+\t-\t-\t%s" % color)
 121         else:
 122             outfile.write("%s\t%s\t%s\t%s" % (fields[1], fields[2], fields[3], fields[0]))
 123             if itemRGB:
 124                 outfile.write("\t1000\t+\t-\t-\t%s" % color)
 125
 126         outfile.write("\n")
 127
 128     outfile.close()
 129
 130
 131 if __name__ == "__main__":
 132     main(sys.argv)