Table 3. Performance of models based on direct feature combination for the bound sequence dataset, evaluated by LOOCV.
Feature | F | SN | SP | ACC | AUC |
A | 0.311 | 0.671 | 0.675 | 0.680 | 0.678 |
A+B | 0.313 | 0.670 | 0.685 | 0.689 | 0.676 |
A+B+C | 0.311 | 0.662 | 0.678 | 0.685 | 0.680 |
A+B+C+D | 0.309 | 0.608 | 0.723 | 0.719 | 0.680 |
A+B+C+D+E | 0.312 | 0.650 | 0.695 | 0.698 | 0.676 |
A+B+C+D+E+F | 0.309 | 0.637 | 0.713 | 0.711 | 0.677 |
A+B+C+D+E+F+G | 0.307 | 0.681 | 0.658 | 0.669 | 0.669 |
A: evolutionary profiles; B: predicted relative accessible surface area; C: physicochemical propensities; D: sparse profile; E: function composition; F: predicted secondary structure; G: amino acid pair.