Martini Mapper: An Automated Fragment-Based Mapping Algorithm for Developing Coarse-Grained Models within the Martini...

Area of Science:

• Computational Chemistry
• Molecular Dynamics Simulations
• Materials Science

Background:

• Coarse-graining (CG) enables large-scale molecular dynamics simulations but faces challenges in creating accurate and transferable models for diverse chemical structures.
• The Martini force field, particularly Martini 3, offers computational efficiency and chemical detail but requires complex mapping procedures for organic molecules.
• Existing automated methods struggle with the complexity and standardization needed for broad chemical applicability.

Purpose of the Study:

• To develop an automated framework for efficiently generating Martini 3 coarse-grained models from Simplified Molecular Input Line Entry System (SMILES) strings.
• To address the limitations in accuracy, transferability, and standardization of CG model construction for diverse organic molecules.
• To enable systematic and scalable generation of Martini 3 models for high-throughput simulations in drug discovery and materials design.

Main Methods:

• Developed an automated framework, Martini Mapper, that combines a curated bead dictionary with a hierarchical, rule-based algorithm.
• The framework generates Martini 3 models directly from SMILES strings, including molecule-specific bonded parameters.
• Mapped 6280 molecules across diverse datasets, including systems requiring bond/angle parameters and large topological mappings.

Main Results:

• Successfully generated Martini 3 models for a large set of diverse molecules, including complex systems beyond the scope of existing automated tools.
• Benchmarking of 1075 mapped structures using transfer free energies showed good agreement with experimental and atomistic reference data.
• Structural validation using Solvent Accessible Surface Area (SASA) further confirmed the accuracy of the generated models.

Conclusions:

• The automated Martini Mapper framework provides a systematic and scalable solution for generating accurate Martini 3 coarse-grained models.
• This approach overcomes previous limitations in model construction efficiency and standardization for a wide range of organic molecules.
• The framework significantly enhances the applicability of Martini 3 simulations for large-scale studies in drug discovery and materials design.