meanstep.py

   1 #!/usr/bin/python
   2
   3 """ meanstep.py
   4  ...
   5 """
   6
   7 import sys
   8 import os
   9
  10 results = {}
  11
  12 if len(sys.argv) != 4 and len(sys.argv) != 5:
  13     print "usage: ./meanstep.py <datadir> <matching col 1>,...,<matching col n> <mean col 1>,...,<mean col n> [<outdir>]"
  14     exit(0)
  15
  16 matchcols = sys.argv[2].split(",")
  17 meancols = sys.argv[3].split(",")
  18
  19 if len(sys.argv) == 5:
  20     outdir = sys.argv[4]
  21     writetofile = True
  22 else:
  23     writetofile = False
  24
  25 for root, dirs, filenames in os.walk(sys.argv[1]):
  26     for f in filenames:
  27         if f[-4:] != ".dat":
  28             continue
  29
  30         if not f[:-4] in results:
  31             results[f[:-4]] = {}
  32             print "Found new fileid: " + f[:-4]
  33
  34         with open(root + "/" + f) as inf:
  35             for line in inf:
  36                 if line[0:1] == "#":
  37                     continue
  38
  39                 linecols = line.split()
  40                 matchkey = ""
  41                 for imatchcol in range(0,len(matchcols)):
  42                     matchkey = matchkey + linecols[int(matchcols[imatchcol])-1]
  43
  44                 if not matchkey in results[f[:-4]]:
  45                     results[f[:-4]][matchkey] = linecols + [1]
  46                 else:
  47                     results[f[:-4]][matchkey][ len(results[f[:-4]][matchkey]) - 1 ] = \
  48                         results[f[:-4]][matchkey][ len(results[f[:-4]][matchkey]) - 1 ] + 1
  49
  50                     for imeancol in range(0,len(meancols)):
  51                         results[f[:-4]][matchkey][int(meancols[imeancol])-1] = \
  52                             float(results[f[:-4]][matchkey][int(meancols[imeancol])-1]) + \
  53                             float(linecols[int(meancols[imeancol])-1])
  54
  55 for fileid in results:
  56     print "\n[" + fileid + "]"
  57
  58     if writetofile:
  59         outfile = open(outdir + "/" + fileid + "_meanstep.dat", "w+")
  60
  61     for matchkey in results[fileid]:
  62         for imeancol in range(0,len(meancols)):
  63             results[fileid][matchkey][int(meancols[imeancol])-1] = \
  64                 float(results[fileid][matchkey][int(meancols[imeancol])-1]) / \
  65                 results[fileid][matchkey][ len(results[f[:-4]][matchkey]) - 1 ]
  66
  67         for icol in range(0,len(results[fileid][matchkey])):
  68             sys.stdout.write(str(results[fileid][matchkey][icol]))
  69             if writetofile:
  70                 outfile.write(str(results[fileid][matchkey][icol]));
  71
  72             if icol != len(results[fileid][matchkey])-1:
  73                 sys.stdout.write("\t")
  74                 if writetofile:
  75                     outfile.write("\t")
  76
  77         print ""
  78         if writetofile:
  79             outfile.write("\n")
  80
  81     if writetofile:
  82         outfile.close()