iSwarm-CG

iSwarm-CG automates bottom-up parametrization of bonded terms in coarse-grained (CG) molecular models by coupling fuzzy self-tuning particle swarm optimization with Boltzmann inversion to reproduce reference all-atom trajectory distributions and generate MARTINI-compatible bonded parameters.

Key Features:

• Automatic iterative parametrization: Automates iterative optimization of bonded parameters in CG models against reference data.
• Fuzzy self-tuning particle swarm optimization (PSO): Employs fuzzy self-tuning PSO to explore parameter space during optimization.
• Boltzmann inversion coupling: Uses Boltzmann inversion together with PSO to derive bonded potentials from reference distributions.
• Bottom-up refinement from all-atom references: Refines bonded parameters to reproduce distributions extracted from reference all-atom trajectories starting from an initial CG model topology and non-bonded parameters.
• MARTINI force field integration: Produces bonded parameters tailored for MARTINI coarse-grained models.
• Gromacs compatibility and solvent evaluation: Compatible with Gromacs for evaluation and optimization in explicit and implicit solvent simulations.
• System size and performance: Handles systems composed of up to 200 pseudo atoms and typically completes parametrization within 4–24 hours on standard desktop machines using default settings.
• Benchmarking: Validated on nine diverse molecules to demonstrate applicability across a range of structural complexities and sizes.

Scientific Applications:

• Coarse-grained model development: Parametrization of bonded terms for MARTINI-based CG models in biomolecular and nanotechnology research.
• Reproduction of all-atom behavior: Refinement of bonded parameters to reproduce structural and conformational distributions from all-atom MD reference trajectories.
• Solvent-dependent parametrization: Optimization of bonded parameters for simulations in explicit and implicit solvent environments using Gromacs.

Methodology:

Couples fuzzy self-tuning particle swarm optimization with Boltzmann inversion to iteratively refine bonded parameters against reference all-atom trajectories, starting from an initial CG model topology and non-bonded parameters.