All you need is water: Converging ligand binding simulations with hydration collective variables

Area of Science:

• Computational chemistry
• Biomolecular simulations

Background:

• Selecting collective variables (CVs) is a major challenge in enhanced sampling molecular dynamics simulations.
• While geometric descriptors are transferable, optimal CVs are system-specific, hindering simulation efficiency.
• Solvation, despite its importance in molecular processes, is under-explored as a CV due to its complexity.

Purpose of the Study:

• To develop a data-driven and automatic strategy for constructing robust solvation-based CVs.
• To demonstrate the effectiveness of these hydration CVs in enhancing sampling and converging binding free energy landscapes.
• To provide guidelines for implementing solvation-based CVs in molecular dynamics simulations.

Main Methods:

• Analyzing the radial distribution function of water around a ligand to identify critical hydration sites.
• Utilizing on-the-fly probability enhanced sampling simulations with the developed hydration CVs.
• Comparing results with computationally expensive benchmark methods for validation.

Main Results:

• Successfully converged binding free energy landscapes for host-guest systems using only hydration CVs.
• Achieved excellent agreement between solvation-based CV results and benchmark methods.
• Identified optimal water biasing strategies for efficient simulation convergence.

Conclusions:

• Solvation-based CVs offer a powerful and generalizable framework for enhancing sampling in complex biomolecular events.
• The developed method highlights the critical role of water in molecular recognition.
• This approach provides practical guidelines for improved molecular dynamics simulations.