Bartender: Martini 3 Bonded Terms via Quantum Mechanics-Based Molecular Dynamics

Area of Science:

• Computational chemistry
• Biophysics
• Materials science

Background:

• Coarse-grained (CG) molecular dynamics (MD) simulations are increasingly vital for studying complex biological and material systems.
• The Martini CG force field (version 3) is widely used but requires manual ligand parametrization, often necessitating computationally expensive all-atom (AA) simulations.
• Existing parametrization methods can yield inaccurate CG models if underlying AA parameters are suboptimal or unknown, limiting simulation accuracy.

Purpose of the Study:

• To introduce Bartender, an automated tool for generating CG models for the Martini 3 force field.
• To leverage quantum mechanics (QM) and MD simulations for efficient and accurate ligand parametrization.
• To enhance the reliability and applicability of CG MD simulations for drug discovery and materials science.

Main Methods:

• Development of Bartender, a novel parametrization tool utilizing QM/MD simulations.
• Implementation in Go for efficient computation and user-friendliness.
• Validation against manually generated models for small molecules and complex ligands.

Main Results:

• Bartender successfully generates accurate bonded terms for Martini 3 CG models.
• For small, cyclic molecules, Bartender-generated models are indistinguishable from human-curated ones.
• For complex, drug-like molecules, Bartender captures dynamical behavior more effectively by fitting advanced functional forms.

Conclusions:

• Bartender offers an efficient and user-friendly solution for Martini 3 ligand parametrization.
• The tool improves the accuracy and numerical stability of CG models, crucial for high-throughput screening.
• Bartender empowers more reliable and predictive CG MD simulations in various scientific domains.