Quality over quantity: Sampling high probability rare events with the weighted ensemble algorithm.

Area of Science:

• Computational chemistry
• Molecular dynamics
• Drug discovery

Background:

• Accurate prediction of molecular binding rates and free energies is crucial for drug design.
• Existing enhanced sampling methods lack a universally reliable workflow for these predictions.

Purpose of the Study:

• To test a modified REVO (Resampling of Ensembles by Variation Optimization) algorithm for enhanced molecular binding free energy calculations.
• To evaluate a new strategy that restricts cloning in quasi-unbound states based on physical criteria.

Main Methods:

• Utilized a modified REVO algorithm, a variant of the weighted ensemble method.
• Applied the modified REVO to retrospectively analyze SAMPL6 host-guest systems and prospectively a SAMPL9 system.
• Implemented a physical criterion (center-of-mass distance) to prevent cloning in states with high unbinding probability.

Main Results:

• The modified REVO algorithm demonstrated improved accuracy or consistency in unbinding free energy calculations across all tested systems.
• The restriction on cloning led to fewer unbinding events, each with a higher statistical weight.
• The new strategy showed enhanced performance compared to the standard REVO algorithm.

Conclusions:

• The modified REVO algorithm represents a significant advancement for calculating transition rates and binding free energies using the weighted ensemble method.
• This flexible approach allows for system-specific criteria to determine cloning eligibility, enhancing predictive power.
• The findings offer a more reliable computational tool for the drug design process.