Fast free energy estimates from λ-dynamics with bias-updated Gibbs sampling.

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• Relative binding free energy calculations are crucial for lead optimization in drug design.
• Classical methods like free energy perturbation and thermodynamic integration are computationally expensive due to independent pairwise perturbations.

Purpose of the Study:

• To develop a more efficient computational method for calculating relative binding free energies.
• To reduce the operational costs and accelerate molecular design workflows.

Main Methods:

• Introduced λ-dynamics with bias-updated Gibbs sampling.
• Employed dynamic biases for collective sampling of multiple ligand analogues within a single simulation.

Main Results:

• Achieved high accuracy in relative binding free energy calculations, with root mean square errors near or below 1.0 kcal/mol for five benchmark systems.
• Demonstrated significant efficiency gains over thermodynamic integration (18-66x for small perturbations, 100-200x for aromatic ring substitutions).
• Results were consistent with other computational approaches and within statistical noise.

Conclusions:

• The λ-dynamics method enables rapid and accurate determination of relative binding free energies.
• This approach accelerates structure-based drug design by allowing exploration of larger chemical spaces.
• The method offers substantial computational savings compared to traditional techniques.