Skip to main content
. 2018 Oct 4;23(10):2535. doi: 10.3390/molecules23102535

Table 2.

Performance comparison of different features on the benchmark dataset (HB34).

Methods Features SPE SEN PRE ACC F1 MCC AUC
SVM Physicochemical 0.672 0.521 0.545 0.608 0.520 0.196 0.566
PSSM 0.696 0.504 0.553 0.614 0.515 0.204 0.634
Blocks substitution matrix 0.644 0.522 0.529 0.594 0.511 0.170 0.595
ASA 0.677 0.688 0.612 0.660 0.638 0.362 0.737
Solvent exposure 0.609 0.726 0.580 0.658 0.635 0.339 0.724
Combined 0.711 0.638 0.684 0.699 0.652 0.393 0.757
RF Physicochemical 0.624 0.549 0.521 0.592 0.522 0.174 0.635
PSSM 0.682 0.561 0.567 0.632 0.555 0.244 0.648
Blocks substitution matrix 0.620 0.550 0.521 0.590 0.523 0.17 0.632
ASA 0.722 0.587 0.614 0.664 0.589 0.312 0.696
Solvent exposure 0.682 0.552 0.565 0.626 0.549 0.236 0.669
Combined 0.756 0.656 0.624 0.699 0.631 0.384 0.766
GTB Physicochemical 0.587 0.586 0.514 0.586 0.535 0.173 0.635
PSSM 0.612 0.641 0.550 0.624 0.584 0.251 0.669
Blocks substitution matrix 0.591 0.588 0.517 0.591 0.540 0.179 0.635
ASA 0.665 0.648 0.588 0.658 0.608 0.310 0.693
Solvent exposure 0.624 0.639 0.558 0.631 0.587 0.261 0.669
Combined 0.717 0.656 0.727 0.719 0.681 0.439 0.787