GROMACS protein–ligand MD simulation workflow
The main stages of a protein–ligand molecular dynamics run in GROMACS, from topology building to trajectory processing.
Why run MD after docking?
Docking gives a static snapshot. Molecular dynamics (MD) simulates the motion of the complex in explicit water over time, showing whether the predicted pose stays bound and how the protein responds.
The workflow
- Prepare the protein topology. Use
gmx pdb2gmxwith a chosen force field and water model. Check chain handling and missing residues first. - Parametrise the ligand. Generate ligand topology and parameters, for example with CGenFF or ACPYPE/GAFF2, and include them in the system topology.
- Build the box and solvate. Define a periodic box with
gmx editconfand add water withgmx solvate. Membrane proteins need a lipid bilayer built first. - Add ions. Neutralise the system and set physiological salt concentration with
gmx genion. - Minimise the energy. Remove bad contacts with steepest descent until the maximum force converges.
- Equilibrate in NVT then NPT. Stabilise temperature, then pressure and density, restraining the ligand and protein backbone so the system relaxes without drifting.
- Run production. Run unrestrained MD for the planned length, for example 50 to 100 ns, writing frames at regular intervals.
- Process the trajectory. Fix periodic boundary effects and centre the complex with
gmx trjconvbefore analysis.
Tips that save time
- Inspect the minimised and equilibrated structure visually before starting production.
- Use checkpoint files so a crashed long run can resume.
- Keep force field, water model and ligand parameters consistent and documented.
- Check temperature, pressure, density and energy plots to confirm equilibration.
Next step
Once the run finishes, analyse stability and interactions. See how to interpret MD simulation results.
Frequently asked questions
How long should a protein–ligand MD simulation be?
Many studies use 50 to 100 ns to assess stability. Larger or more flexible systems and free-energy work can need several hundred nanoseconds.
Which force field should I use?
CHARMM36m and AMBER ff14SB or ff19SB are widely used for proteins. Ligand parameters usually come from CGenFF with CHARMM or GAFF2 with AMBER. Keep the families matched.
Do I need a GPU?
Not required, but a GPU cuts production run time dramatically for tens to hundreds of nanoseconds.
What is the difference between NVT and NPT?
NVT stabilises temperature at constant volume. NPT then stabilises pressure and density at constant temperature before production.
Related guides
Look for convergence, not single numbers. Combine several metrics before concluding that a ligand is stably bound.
Free-energy guideMM/PBSA Binding Free Energy Explained for BeginnersMM/PBSA estimates binding energy from MD snapshots. It ranks similar ligands reasonably well but is not an exact measure of affinity.
Related service
Written by Sohel Akhtar Ansari, ReactorMD founder (ORCID). Computational predictions are not experimental confirmation.