BAR-based optimum adaptive sampling regime for variance minimization in alchemical transformation

Area of Science:

• Computational Chemistry
• Molecular Dynamics
• Free Energy Simulations

Background:

• Alchemical free energy simulations are crucial for predicting molecular properties.
• Current methods like standard BAR often use uniform sampling, potentially leading to suboptimal efficiency.
• Adaptive strategies are needed to optimize the weighting of simulation ensembles.

Purpose of the Study:

• To improve the efficiency and accuracy of alchemical free energy simulations.
• To introduce a novel adaptive method for manipulating ensemble significance.
• To minimize the total variance in free energy estimates.

Main Methods:

• Development of the Optimum Benefit-Risk Ratio (OBAR) method.
• Adaptive manipulation of ensemble significance using importance sampling.
• Proposal of the Time Derivative of total Variance (TDV) as an OBAR criterion.
• Application to solvation and protein-ligand binding systems.

Main Results:

• OBAR method explicitly considers statistical inefficiency, outperforming the traditional equal time rule.
• The TDV criterion is more sensitive to importance rank than the overlap matrix.
• Demonstrated improved performance for solvation of small molecules.
• Successfully applied to a complex protein-ligand binding system.

Conclusions:

• The OBAR method offers a more statistically robust approach to alchemical free energy calculations.
• Adaptive ensemble weighting via TDV enhances the precision of free energy predictions.
• OBAR represents a significant advancement for computational molecular modeling.