makeSNPtrack.py

   1 #
   2 #  makeSNPtrack.py
   3 #  ENRAGE
   4 #
   5 # This script maps all the qualified SNC sites on to the genome browser
   6 # Output format: bed
   7 # Written by: Wendy Lee
   8 # Written on: August 18th, 2008
   9 # Last Modified: December 14th, 2008 by Ali Mortazavi
  10
  11 import sys
  12
  13 def main(argv=None):
  14     if not argv:
  15         argv = sys.argv
  16
  17     print "makeSNPtrack: version 1.3"
  18
  19     if len(argv) < 4:
  20         print "usage: python %s snpfile trackname trackoutfile" % argv[0]
  21         sys.exit(1)
  22
  23     snpfile = argv[1]
  24     track = argv[2]
  25     outfile = argv[3]
  26
  27     makeSNPtrack(snpfile, track, outfile)
  28
  29
  30 def makeSNPtrack(snpfilename, track, outfilename):
  31
  32     snpfile = open(snpfilename, "r")
  33     writeSNPsBedfile(snpfile, track, outfilename)
  34     snpfile.close()
  35
  36
  37 def writeSNPsBedfile(snpPropertiesList, track, outfilename):
  38
  39     outfile = open(outfilename, "w")
  40     header = getHeader(track)
  41     outfile.write(header)
  42
  43     for line in snpPropertiesList:
  44         if doNotProcessLine(line):
  45             continue
  46
  47         fields = line.strip().split()
  48         outline = getBedOutputLine(fields)
  49         outfile.write(outline)
  50
  51     outfile.close()
  52
  53
  54 def getHeader(track):
  55     header = "track name=%s description=%s visibility=2 itemRgb=\"On\"\n" % (track, track)
  56     return header
  57
  58
  59 def doNotProcessLine(line):
  60     return line[0] == "#"
  61
  62
  63 def getBedOutputLine(snpPropertiesList):
  64     chromosome = snpPropertiesList[2]
  65     readStart = int(snpPropertiesList[3]) - 1
  66     readStop = readStart + 1
  67     readName = snpPropertiesList[7]
  68     color = getSNPColor(readName)
  69     score = "0"
  70     sense = "+"
  71     outline = "%s\t%d\t%d\t%s\t%s\t%s\t-\t-\t\t%s\n" % (chromosome, readStart, readStop, readName, score, sense, color)
  72
  73     return outline
  74
  75
  76 def getSNPColor(readName):
  77     baseColor = {"A": "200, 0, 255",
  78                  "T": "200, 0, 255",
  79                  "C": "200, 0, 255",
  80                  "G": "200, 0, 255"
  81     }
  82
  83     specialColors = {"A-G": "255, 0, 0",
  84                      "T-C": "0, 0, 255"
  85     }
  86
  87     if readName in specialColors.keys():
  88         color = specialColors[readName]
  89     else:
  90         try:
  91             color = baseColor[readName[-1]]
  92         except (IndexError, KeyError):
  93             color = "200, 0, 255"
  94
  95     return color
  96
  97
  98 if __name__ == "__main__":
  99     main(sys.argv)