Table 3. Goodness-of-fit (R2) and predictive ability (Q2 test) of the obtained models.
Models with different descriptor combinations | GD | DLI | ||
R2 | Q2 test | R2 | Q2 test | |
PLIFa | 0.9621 | 0.7470 | 0.9621 | 0.7470 |
MLPDa | 0.9700 | 0.7101 | 0.9722 | 0.6702 |
L & P & PLIFb | 0.9716 | 0.8271 | 0.9731 | 0.7929 |
L & P &MLPDb | 0.9696 | 0.7129 | 0.9727 | 0.6612 |
L&Pc | 0.9350 | 0.7298 | 0.9241 | 0.6134 |
Models created using only cross-terms.
Models created using ligand and protein descriptors with cross-terms.
Models created using ligand and protein descriptors.