tools/evaluatePrediction.py

   1 #!/usr/bin/env python
   2 # -*- coding:latin-1 -*-
   3
   4 import sys
   5 import cPickle
   6
   7 ground_truth   = cPickle.load(sys.argv[1])
   8 prediction     = cPickle.load(sys.argv[2])
   9
  10 ESTs     = ground_truth['TestEsts']
  11 Exons    = ground_truth['TestExon']
  12 Acc      = ground_truth['TestAcc']
  13 Don      = ground_truth['TestDon']
  14 DNA      = ground_truth['Test']
  15
  16 totalWrongPos = [0]*len(prediction)
  17
  18 for pos,alignmentObj in enumerate(prediction):
  19    predPos  = zips(alignmentObj.tStarts, alignmentObj.tEnds)
  20    wrongPos  = 0
  21
  22    for predExonStart,predExonStop in predPos:
  23
  24       for pos,currentExons in enumerate(Exons):
  25          for idx in range(1,len(currentExons)-1):
  26
  27             intronStart = currentExons[idx][1]
  28             intronStop  = currentExons[idx+1][0]
  29
  30             # est is covering full intron
  31             if intronStart > predExonStart and intronStop < predExonStop:
  32                #fprintf(fh,sprintf(' %s is completely overlapping from %d %d\n',currentESTName,intronStart,intronStop));
  33                wrongPos += intronStop-intronStart
  34             # est is nested inside intron
  35             elif intronStart < predExonStart and intronStop > predExonStop:
  36                #fprintf(fh,sprintf(' %s is completely inside intron from %d %d\n',currentESTName,predExonStart,predExonStop));
  37                wrongPos += intronStop-intronStart
  38             # end of exonth is nested inside predExoniction
  39             elif intronStart > predExonStart and predExonStop > intronStart and intronStop > predExonStop:
  40                #fprintf(fh,sprintf('%s is upstream overlapping from %d %d\n',currentESTName,intronStart,predExonStop));
  41                wrongPos += intronStop-intronStart
  42             # predExoniction is nested inside exonth
  43             elif intronStart < predExonStart and predExonStart < intronStop and intronStop < predExonStop:
  44                #fprintf(fh,sprintf('%s is downstream overlapping from %d %d\n',currentESTName,predExonStart,intronStop));
  45                wrongPos += intronStop-intronStart
  46             else:
  47                pass
  48
  49    totalWrongPositions[pos] = wrongPos