Swarm-CG

Swarm-CG automates the parametrization of bonded terms in coarse-grained (CG) molecular models to derive bonded parameters consistent with all-atom (AA) reference trajectories.

Key Features:

• Automatic Optimization: Automates optimization of bonded parameters in CG models starting from a preliminary model that includes topology and non-bonded parameters.
• Metaheuristic Algorithm: Implements fuzzy self-tuning particle swarm optimization for automated parameter search.
• Boltzmann Inversion: Couples metaheuristic optimization with Boltzmann inversion to extract bonded distributions from reference data.
• Reference Data: Uses all-atom (AA) reference trajectories to guide parametrization of bonded terms.
• Forcefield Compatibility: Compatible with the MARTINI forcefield and applicable to other coarse-grained schemes.
• Scalability and Performance: Handles models up to 200 pseudoatoms and completes parametrization in approximately 4–24 hours on standard desktop hardware.
• Benchmark Validation: Validated on a benchmark set of nine diverse molecules.

Scientific Applications:

• Coarse-Grained Model Development: Generation and refinement of bonded parameters for new CG models using AA trajectory references.
• Biotechnology Research: Parameter refinement for CG simulations applied to biological systems in bio-technology studies.
• Nanotechnology and Nanostructure Design: Parametrization of CG models used to study and design novel nanostructures.

Methodology:

Swarm-CG couples fuzzy self-tuning particle swarm optimization with Boltzmann inversion, starting from a preliminary CG model (topology and non-bonded parameters) and using all-atom (AA) reference trajectories to fit bonded terms.