Enhancing Ligand and Protein Sampling Using Sequential Monte Carlo.

Area of Science:

• Computational Chemistry
• Computational Biophysics
• Molecular Dynamics

Background:

• The sampling problem is a critical challenge in computational chemistry, requiring efficient methods to explore molecular configurations.
• Existing sampling techniques often necessitate system-dependent hyperparameters, limiting their broad applicability and efficiency.

Purpose of the Study:

• To present and evaluate an alchemical variation of adaptive sequential Monte Carlo (SMC) for enhanced molecular sampling.
• To demonstrate the versatility and efficiency of alchemical SMC across diverse computational biophysics systems.

Main Methods:

• Developed an alchemical adaptation of sequential Monte Carlo (SMC), an irreversible importance resampling method.
• Applied alchemical SMC to various test cases: ligand torsional rotations, protein-ligand binding, and protein side-chain dynamics.

Main Results:

• Alchemical SMC demonstrated efficient exploration of targeted molecular degrees of freedom.
• Consistent performance was achieved across different systems using uniform hyperparameters.
• The method proved effective for systems requiring trade-offs between long-timescale and short-timescale sampling.

Conclusions:

• Alchemical SMC is a powerful and broadly applicable tool for molecular sampling in computational biophysics.
• It offers a promising approach for preparatory system exploration, optimizing sampling efficiency.
• The method's adaptability and consistent performance reduce the need for system-specific parameterization.