Estimating the ligand-binding affinity via λ-dependent umbrella sampling simulations.

Abstract

The umbrella sampling (US) approach has been demonstrated to be a very efficient method for estimating the ligand-binding affinity. However, most of the calculated values overestimate experimental ones that are probably caused by the inaccurate representation of the interaction between the ligand and the surrounding molecules. The issue can be resolved via the implementation aspects of λ-alteration simulation into the US approach, which we call the λ-dependent umbrella sampling (λUS) scheme. In particular, the electrostatic and van der Waals interactions were simultaneously changed by using the coupling parameter λ during λUS simulations. The mean value of obtained results, kcal mol-1 , is in good fitting to the mean value of respective experiments, ∆GEXP = - 11.26 ± 0.89 kcal mol-1 . Moreover, the correlation between the proposed approach and experiment is quite good with a value of . The λUS scheme significantly enhances the calculated accuracy since the RMSE of the proposed scheme is smaller than traditional US simulations, kcal mol-1 versus kcal mol-1 . Furthermore, the precision is increased since the computed error via λUS approach, kcal mol-1 , was smaller than those of the US simulation, kcal mol-1 . Overall, the proposed approach perhaps provides an efficient way to accurately and precisely estimate the ligand-binding free energy.