How to Use Martini OpenMM Molecular Dynamics

Martini OpenMM Molecular Dynamics is a coarse-grained simulation workflow for researchers who want to explore longer-timescale protein behavior with a reduced-resolution model. Instead of representing every atom explicitly, Martini groups atoms into beads, which can make membrane organization, large protein motions, and solvent-scale dynamics more accessible than all-atom simulations at the same compute budget.

On Neurosnap, the workflow starts from a protein-only Input Structure and optional Input Small Molecules supplied as positioned SDF files. The service coarse grains the protein with martinize2, can generate a Martini membrane with insane, parameterizes supported small organic ligands with Auto_MartiniM3 for Martini 3 jobs, and runs the resulting system with CUDA-backed OpenMM.

The current automated workflow is intended for proteins, small organic ligands, solvent, ions, and membranes. It does not currently support protein/nucleic-acid complexes, embedded PDB HETATMs, cofactors, waters, or ions in the input structure; those environment components are generated during setup.

A typical Martini OpenMM workflow includes structure cleanup, protein coarse graining, optional ligand coarse graining, Martini system construction, energy minimization, NVT equilibration, NPT equilibration, and production dynamics. martinize2 converts the protein to Martini beads and topology files, Auto_MartiniM3 provides best-effort Martini 3 ligand parameters for small organic SDF ligands, and insane builds solvent or membrane environments before OpenMM runs the simulation.

On Neurosnap, Martini Force Field, Solvent, Salt Concentration, Add Membrane, box dimensions, elastic-network settings, timestep, and equilibration controls define the physical model and startup protocol. Elastic Network can improve coarse-grained structural stability but may substantially increase martinize2 setup time for larger or multichain proteins. Output Frames controls production trajectory sampling and is bounded adaptively to avoid excessive output size. Ligand SDF files must include usable 3D coordinates already positioned near the intended protein site; the service preserves that placement through the coarse-graining workflow.

Results should be interpreted as coarse-grained bead trajectories rather than atomistic structures. The final PDB, multi-model production PDB, DCD trajectory, state-data plots, serialized system.xml, and martini_inputs.zip support review of system stability, temperature control, pressure-coupled volume or density behavior, and downstream reproduction in Martini-compatible tooling.

• Longer-timescale coarse graining: Martini reduces molecular detail into beads, making large protein and membrane simulations more practical than equivalent all-atom runs.
• Automated setup path: Neurosnap chains protein coarse graining, optional Auto_MartiniM3 ligand parameterization, solvent or membrane construction, equilibration, and production in one workflow.
• Membrane-ready controls: Lipid selection, box dimensions, salt concentration, pressure coupling, and elastic-network settings expose the key Martini setup decisions without manual topology editing.
• Reproducible outputs: Trajectories, state data, serialized OpenMM systems, generated Martini topology files, and processed input bundles make each coarse-grained run inspectable and portable.