Skip to main content

Table 3 Performance evaluation (auROC and auPRC scores) of four different methods on a subset of the genome-wide cress and human datasets.

From: Accurate splice site prediction using support vector machines

  SM WD WDSP WDS WDSSP
Cress      
   Acceptor      
   auROC 99.41 98.97 99.36 99.43 99.43
   auPRC 91.76 84.24 90.64 92.01 92.09
   Donor      
   auROC 99.59 99.38 99.58 99.61 99.61
   auPRC 93.34 88.62 93.42 93.68 93.87
Human      
   Acceptor      
   auROC 97.72 97.34 97.71 97.73 97.82
   auPRC 50.39 42.77 50.48 51.78 54.12
   Donor      
   auROC 98.44 98.36 98.36 98.51 98.37
   auPRC 53.29 46.53 54.06 53.08 54.69
  1. The methods compared are SpliceMachine (SM), the weighted degree kernel (WD), the weighted degree kernel complemented with six spectrum kernels (WDSP), the weighted degree kernel with shifts (WDS), and the weighted degree kernel with shifts complemented by six spectrum kernels (WDSSP).