Table 4. The MS-QM/MM-LR Correlations (eq 5) for Different Tautomer and Species Composition: Optimized Coefficients and Statistical Indices.
LOOc |
MC-LGOd |
|||||||||
---|---|---|---|---|---|---|---|---|---|---|
tautomer | species | α × 10–3(mol/kcal)a | γ × 10–3(1/Å2)b | κ | r2 | RMSE | q2 | RMSE | q2 | RMSE |
single | single | –1.212 ± 1.289 | –8.985 ± 8.451 | 1.151 ± 0.896 | 0.662 | 0.431 | 0.617 | 0.458 | 0.616 | 0.459 |
single | multi | –1.291 ± 1.026 | –8.664 ± 7.211 | 1.000 ± 0.758 | 0.734 | 0.382 | 0.690 | 0.412 | 0.691 | 0.411 |
multi | single | –1.338 ± 0.801 | –8.334 ± 5.002 | 0.960 ± 0.507 | 0.839 | 0.297 | 0.819 | 0.315 | 0.819 | 0.315 |
multi | multi | –1.326 ± 0.591 | –7.761 ± 2.261 | 1.160 ± 0.358 | 0.906 | 0.228 | 0.897 | 0.237 | 0.894 | 0.241 |
Scales the QM/MM energy term.
Scales the SASA term.
Leave-one-out cross-validation: the results reported as RMSE and the squared predictive correlation coefficient (q2).
Monte Carlo leave-group-out cross-validation: the compounds were randomly divided into 7 groups, and each group of ∼9 compounds was used once as a test set; the procedure was repeated 10 times.