projects/recombineMap/sperm_by_fw/crossover_prediction/HMM_method/compareData.py

   1 '''
   2 Created on 2012-7-12
   3
   4 @author: LiJinsen
   5 '''
   6 import sys, os
   7
   8 filename = sys.argv[1]
   9 filename2 = sys.argv[2]
  10
  11 #raw data
  12 file = open(filename,'r')
  13 content1 = file.read()
  14 result1 = {}
  15 for line in content1.split('\n'):
  16     if len(line)<1:
  17         continue
  18     item = line.split('\t')
  19     if not result1.has_key(item[0]):
  20         result1.update({item[0]:{}})
  21     if not result1[item[0]].has_key(item[1]):
  22         result1[item[0]].update({item[1]:[0]})
  23     result1[item[0]][item[1]][0]+=1
  24     result1[item[0]][item[1]].append(item)
  25
  26 #HMM data
  27 file = open(filename2,'r')
  28 content2 = file.read()
  29 result2 = {}
  30 for line in content2.split('\n'):
  31     if len(line)<1:
  32         continue
  33     item = line.split('\t')
  34     if not result2.has_key(item[0]):
  35         result2.update({item[0]:{}})
  36     if not result2[item[0]].has_key(item[1]):
  37         result2[item[0]].update({item[1]:[0]})
  38     result2[item[0]][item[1]][0]+=1
  39     result2[item[0]][item[1]].append(item)
  40
  41 #print result1
  42
  43 w = open('difference.txt','w')
  44 w1 = open('differenceDetail.raw.txt','w')
  45 w2 = open('differenceDetail.new.txt','w')
  46 def writeData(item1, item2):
  47     for obj in item1:
  48         flag = False
  49         for tmpobj in item2:
  50             if obj[2]==tmpobj[2]:
  51                 flag = True
  52                 break
  53         if flag==True:
  54             continue
  55         for c in obj:
  56             w2.write(c)
  57             w2.write('\t')
  58         w2.write('\n')
  59
  60 for itemkey1 in result1.keys():
  61     for itemkey2 in result1[itemkey1].keys():
  62         if not result2[itemkey1].has_key(itemkey2):
  63             w.write(itemkey1)
  64             w.write('\t')
  65             w.write(itemkey2)
  66             w.write('\tNot In\n')
  67             continue
  68         if result1[itemkey1][itemkey2][0] != result2[itemkey1][itemkey2][0]:
  69             w.write(itemkey1)
  70             w.write('\t')
  71             w.write(itemkey2)
  72             w.write('\t')
  73             if result1[itemkey1][itemkey2] < result2[itemkey1][itemkey2]:
  74                 w.write('More')
  75             else:
  76                 w.write('Less')
  77             w.write('\n')
  78             writeData(result2[itemkey1][itemkey2][1:], result1[itemkey1][itemkey2][1:])
  79