Skip to main content
. 2021 Jan 25;17(2):1250–1265. doi: 10.1021/acs.jctc.0c01179

Table 1. Validation of the TIES 20 Implementation along with the Previous Results (TIES 17) with Respect to the Experimental ΔΔGa.

protein property TIES 17 TIES 20 TIES 20E TIES 20 BCC
TYK2 MUE 0.44 (0.14) 0.48 (0.14) 0.43 (0.14) 0.66 (0.19)
MSE –0.03 (0.24) –0.00 (0.24) 0.00 (0.23) –0.02 (0.31)
RMSE 0.56 (0.17) 0.58 (0.16) 0.53 (0.16) 0.86 (0.22)
Pearson’s ρ 0.94 0.93 0.94 0.84
slope 0.83 0.86 0.84 0.66
intercept 0.12 0.07 0.08 0.20
MCL1 MUE 1.20 (0.23) 1.08 (0.23) 1.09 (0.22) 0.81 (0.17)
MSE 0.38 (0.38) 0.09 (0.38) 0.18 (0.37) 0.22 (0.28)
RMSE 1.41 (0.26) 1.34 (0.27) 1.32 (0.26) 0.98 (0.18)
Pearson’s ρ 0.80 0.68 0.70 0.83
slope 1.18 0.79 0.83 0.93
intercept –0.32 –0.16 –0.23 –0.24
thrombin MUE 0.71 (0.17) 0.54 (0.17) 0.53 (0.17) 0.52 (0.15)
MSE 0.19 (0.29) –0.04 (0.28) –0.12 (0.27) –0.04 (0.26)
RMSE 0.78 (0.18) 0.65 (0.20) 0.67 (0.19) 0.63 (0.17)
Pearson’s ρ 0.91 0.87 0.86 0.90
slope 1.50 1.20 1.20 1.29
intercept –0.07 0.09 0.17 0.11
PTP1B MUE 0.37 (0.15) 0.68 (0.22) 0.69 (0.22) 0.66 (0.21)
MSE 0.11 (0.25) –0.24 (0.36) –0.25 (0.35) –0.54 (0.30)
RMSE 0.48 (0.17) 0.83 (0.26) 0.84 (0.26) 0.83 (0.26)
Pearson’s ρ 0.84 0.42 0.41 0.66
slope 0.86 0.30 0.31 0.53
intercept –0.17 –0.08 –0.07 0.33
CDK2 MUE 0.74 (0.21) 0.86 (0.27) 0.85 (0.26) 0.61 (0.24)
MSE 0.17 (0.38) 0.14 (0.46) 0.21 (0.44) 0.02 (0.38)
RMSE 0.81 (0.21) 1.06 (0.29) 1.02 (0.27) 0.83 (0.30)
Pearson’s ρ 0.87 0.75 0.76 0.86
slope 1.28 1.15 1.12 1.28
intercept –0.10 –0.10 –0.18 0.05
a

In TIES 20E, sampling is enhanced by increasing the number of replicas per λ window to 20. All TIES use RESP charges except for TIES 20 BCC which uses the AM1-BCC charges. Due to the small number of cases in PTP1B, one outlier was removed as it was skewing the statistics. The original data can be found in Table S1. The differences (MUE, MSE, RMSE) between calculated and experimental binding free energies are calculated by resampling the data from simulations and experiments (see the Methods section). The uncertainties of these metrics (given in parentheses) are the standard deviations of the resampling distributions. All energies are in kcal/mol.