Lambda-ABF: Simplified, Portable, Accurate and Cost-effective Alchemical Free Energy Computations

TL;DR

Lambda-ABF is a novel, efficient method for alchemical free energy calculations that combines multiple walker adaptive biasing with lambda-dynamics, reducing computational cost and requiring minimal user input.

Contribution

The paper introduces Lambda-ABF, a new approach that simplifies and improves alchemical free energy computations by eliminating manual lambda schedule optimization and enabling unbiased, cost-effective results.

Findings

Accurately computes free energies with less computational cost.

Achieves results within 1 kcal/mol of experimental data.

Implemented in NAMD and Tinker-HP for practical use.

Abstract

We introduce an efficient and robust method to compute alchemical free energy differences, resulting from the application of multiple walker Adaptive Biasing Force (ABF) in conjunction with strongly damped Langevin $\lambda$-dynamics. Unbiased alchemical free energy surfaces are naturally recovered by Thermodynamic Integration (TI). No manual optimization of the $\lambda$ schedule is required as the sampling of the $\lambda$ variable is continuous and converges towards a uniform distribution. Free diffusion of $\lambda$ improves orthogonal relaxation compared to fixed $\lambda$ methods such as standard TI or Free Energy Perturbation (FEP). Furthermore, the multiple walker strategy provides coverage of orthogonal space in a generic way with minimal user input and negligible computational overhead. Of practical importance, no adiabatic decoupling between the alchemical and Cartesian degrees of freedom is assumed, ensuring unbiased estimates for a wide envelope of numerical parameters. We present two high-performance implementations of the method in production molecular dynamics engines, namely NAMD and Tinker-HP, through coupling with the Colvars open source library. These interfaces enable the combination of the rich feature sets of those packages. We demonstrate the correctness and efficiency of the approach on several real-world cases: from solvation free energies up to ligand-receptor binding (using a recently proposed binding restraint scheme) with both fixed-charge and polarizable models. We find that, for a chosen accuracy, the computational cost is strongly reduced compared to state-of-the-art fixed-lambda methods and that results within 1~kcal/mol of experimental value are recovered for the most complex system. The implementation is publicly available and readily usable by practitioners of current alchemical methods.