Lessons Learned from a Ligand-Unbinding Stress Test for Weighted Ensemble Simulations.

Area of Science:

• Computational chemistry
• Biophysics
• Molecular dynamics

Background:

• The weighted ensemble (WE) path sampling strategy enables molecular simulations of biological processes beyond millisecond timescales.
• Further optimization of WE is needed to fully realize its potential, particularly through rigorous stress testing.

Purpose of the Study:

• To stress test the WE method by simulating the seconds-timescale unbinding of ADP from the Eg5 motor protein.
• To identify improvements for the WE resampling procedure and progress coordinate selection.
• To develop a novel, more efficient WE method.

Main Methods:

• Exploration of a seconds-timescale ligand-protein unbinding event using WE path sampling.
• Development of the minimal adaptive binless (MABL) WE method, inspired by minimal adaptive binning.
• Comparison of MABL's efficiency against binned WE approaches.

Main Results:

• The stress test provided insights into optimizing progress coordinates and WE resampling.
• The minimal adaptive binless (MABL) method was developed, offering a binless alternative to rectilinear binning.
• MABL demonstrated >50% greater efficiency compared to its binned counterpart.

Conclusions:

• The MABL method represents a significant advancement in WE path sampling, enhancing efficiency for complex molecular simulations.
• This binless approach provides a foundation for developing more sophisticated WE methods.
• The study highlights the value of stress testing in advancing computational simulation strategies.