Thermostat and barostat in equilibration phase in a membrane-protein system

GROMACS version: gromacs-2022
GROMACS modification: Yes/No


I am using CHARMM-GUI generated system of a transporter protein in a membrane bilayer. In the equilibration phase, the recommended workflow from CHARMM-GUI is to use Berendsen for both temperature and pressure (semi-isotropic). However, gromacs highly discourage it, giving warnings for its usage:

"Command line:
gmx_mpi grompp -f step6.3_equilibration.mdp -o step6.3_equilibration.tpr -c step6.2_equilibration.gro -r step5_input.gro -p -n index.ndx -maxwarn 1

Replacing old mdp entry ‘nstxtcout’ by ‘nstxout-compressed’

WARNING 1 [file step6.3_equilibration.mdp]:
The Berendsen thermostat does not generate the correct kinetic energy
distribution, and should not be used for new production simulations (in
our opinion). We would recommend the V-rescale thermostat.

WARNING 2 [file step6.3_equilibration.mdp]:
The Berendsen barostat does not generate any strictly correct ensemble,
and should not be used for new production simulations (in our opinion).
For isotropic scaling we would recommend the C-rescale barostat that also
ensures fast relaxation without oscillations, and for anisotropic scaling
you likely want to use the Parrinello-Rahman barostat.

Generating 1-4 interactions: fudge = 1
Number of degrees of freedom in T-Coupling group SOLU is 8249.69
Number of degrees of freedom in T-Coupling group MEMB is 71878.31
Number of degrees of freedom in T-Coupling group SOLV is 73323.00

NOTE 1 [file step6.3_equilibration.mdp]:
Removing center of mass motion in the presence of position restraints
might cause artifacts. When you are using position restraints to
equilibrate a macro-molecule, the artifacts are usually negligible.

There was 1 note

There were 2 warnings"

My question here is if its safe to maxwarn 2 in this case. I have actually used v-rescale as a thermostat and berendsen as a barostat to bypass it (just for equilibration), but recently I have discovered that its highly not recommended as well. In addition, if I am not mistaken, when I tried to use Parrinello-Rahman, it didn’t work with GPUs (but I need to check this one again).
I am not sure what is the right approach here: To use other barostat without GPU processing? Ignore warnings because as long as it is not production, it is tolerable? How about v-rescale and c-rescale combination?

Thank you,


I think that it is a good option. It is always better to avoid ignore warnings

Thanks for your reply, I think I will do that.
However, I still have concerns because CHARMM-GUI suggests berendsen in a membrane environment (although I think it suggests the same in a solution builder as well). Perhaps as long as the end protein is stable and whole at the end of equilibration then it doesn’t matter?

The CHARMM-GUI protocol was established a number of years ago and is considered robust. In the intervening time, GROMACS has elevated the level of notification about Berendsen methods from a Note to a Warning, which now causes a problem. Please provide feedback to the CHARMM-GUI developers via their website, and they may consider an update.

Then it is fine to “maxwarn -2” (in equilibration phase only of course) and it is just gromacs definitions? I will search if CHARMM-GUI has forums or support section for these questions. Thanks for the reply !

Or just switch the settings to better thermostat and barostat algorithms. There’s no reason to necessarily think that the Berendsen methods are superior to V-rescale and C-rescale these days.

I understand, thanks to you both!

Dear sir,
Is C-rescale is appropriate also for production MD runs? Does it have advantages over some classic 2nd order thermostats as Parrinello Rahman in production MD runs?