Parrinelo-rahman parameters and thermostat for lipid membranes using charmm C36m

GROMACS version: 2021.5
GROMACS modification: No
Force field: Charmm C36m 2022

Hello, I’m performing a simulation of an antimicrobial peptide with a few membrane models, I generated my membranes with charmm-gui, and I’m going to perform a production ran of 1µs with the membranes alone before inserting the peptide into the box. I’ve been searching the mdp parameters for charmm c36m in the literature, but the parameters for the thermostat and barostat are all over the place, most authors use the Nosé−Hoover thermostat for membranes and the paper on the validation of the c36 forcefield for lipids uses the Nosé−Hoover−Langevin piston for barostat, which is not available in gromacs, on top of that the C36m paper cites many different combinations of thermostats and barostats for peptides. Finally, I’m also not sure about what value of tau_p to use with the Parinello-rahman barostat, the protein-ligand (which also uses charmm) tutorial uses uses a tau_p of 10.0 ps and the V-rescale thermostat. But the mdp files generated by charmm-gui use a tau_p of 5.0 and the Nosé-Hoover thermostat. I don’t trust the charmm-gui mdps though as they are full of outdated entries and missing others. Many people on zenodo seem to use these mdps for their simulations though, they don’t even rename them.

Any help is appreciated

My production mdp file can be found below

md.mdp (2.0 KB)

Thermostat and barostat parameters are not part of a force field, but are part of the MD protocol itself.

It is okay to question the content of auto-generated input files instead of just using them blindly, but in this case a factor of two in the time constant should not be a big difference. You want to be operating in a regime where there are hardly any differences with respect to tau_p, i.e. you want to have a system that is structurally stable, including fluctuations in volume and area.

If your initial configuration is reasonable, you could use any of the thermostats/barostats mentioned that give stable behavior. If not, the Berendsen schemes are good to reach equilibrium from afar (but not for sampling the correct fluctuations). If you know what canonical sampling is, you will find plenty of literature to that sense.

If not, feel free to compare with other atomistic force field, as they are similar in order of magnitude of the forces, which is what matters for the barostats.

Giacomo

Thank you for your reply, it was quite helpful!!