Skip to main content

Table 6 Predictive performance based on selected features and machine learning algorithms based on validation sets and test sets

From: Prediction of HIV-1 protease cleavage site using a combination of sequence, structural, and physicochemical features

Datasets Features Algorithm Sen.(%) Spe.(%) Acc.(%) AUC
746 Seq + Str ANN 100.0 (100.0)* 88.2 (94.4) 94.5 (97.4) 0.994 (0.995)
1625 Seq + Str ANN 94.7 (89.4) 95.2 (96.8) 95.1 (95.1) 0.992 (0.994)
Schilling Seq + Str + Phy ANN 57.1 (27.3) 96.5 (95.8) 91.4 (86.6) 0.895 (0.815)
Impens SA ANN 71.4 (44.4) 100.0 (89.8) 95.7 (80.0) 0.950 (0.816)
  1. *The predictive performance of test set is shown in parenthesis