Skip to main content

Table 1 Individual classification performances of FGA-generated feature sets for 3' (A-KmerX) and 5' (D-KmerX) splice sites.

From: Features generated for computational splice-site prediction correspond to functional elements

A-3mer 86.46 A-4mer 84.92
A-3mer1 84.16 A-4mer1 77.28
A-3mer2 77.01 A-4mer2 69.10
A-3mer3 69.42 A-4mer3 63.11
A-3mer4 63.30 A-4mer4 56.66
A-3mer5 56.84 A-4mer5 49.23
A-3mer6 49.50 A-4mer6 41.02
A-3mer7 41.22   
A-5mer 80.60 A-6mer 68.64
A-5mer1 69.20 A-6mer1 61.72
A-5mer2 62.74 A-6mer2 54.65
A-5mer3 56.25 A-6mer3 47.19
A-5mer4 49.08 A-6mer4 39.62
A-5mer5 40.51   
D-3mer 86.79 D-4mer 85.21
D-3mer1 83.45 D-4mer1 81.14
D-3mer2 80.31 D-4mer2 70.47
D-3mer3 70.08 D-4mer3 55.38
D-3mer4 56.06 D-4mer4 44.77
D-3mer5 42.97   
D-5mer 83.64 D-6mer 75.03
D-5mer1 77.20 D-6mer1 66.68
D-5mer2 57.42 D-6mer2 43.31
D-5mer3 38.09   
  1. FGA-generated feature sets for splice sites and their individual performances at splice-site prediction. Each value reported is an average precision (positive predictive value, TP/(TP+FP)) over 11 values of recall (sensitivity, TP/(TP+FN)), 0%, 10%, 20% ... and 100%, and is the result of a three-fold cross validation. All the features in these features sets extend along the whole splice-site neighbourhood [-82, 80] that we study.