Optimal MDP for Gromos 54a7 ff in latest Gromacs versions

User discussions
1

GROMACS version:
GROMACS modification: 2023.3

Can you suggest optimal parametrs to set in .mdp file (on all or some stages as minimization, nvt, npt and md) for Gromos 54a7 ff?
I’m using 2023.3 version.

After looking through other posts I found something like that, but I’m not still confindent with this choice:
constraint_algorithm = lincs
constraints = all-bonds (or may be h-bonds)
lincs_iter = 1
lincs_order = 4
cutoff-scheme = Verlet
nstlist = 5 (or may be 10)
ns_type = grid
rlist = 1.4 (or may be 1.0)
coulombtype = PME
rcoulomb = 1.4 (or may be 1.0)
vdw-type = Cut-off
rvdw = 1.4
fourierspacing = 0.12 (or may be 0.18)
pbc = xyz
DispCorr = EnerPres (or may be no)

2

I should first note that there is, or as least has been, disagreement in the GROMOS community over whether GROMACS, or any other package than GROMOS for that matter, can properly use the GROMOS force fields. These have been parametrized with an incorrect, non energy conserving, twin range scheme. Also some people in the GROMOS community still seem to think that PME is a bad idea. With that said, we have had good results with a correct, symplectic twin-range multiple time stepping setup. Now GROMACS no longer supports twin-range cut-offs, so you need to use a single 1.4 nm cut-off. This is more accurate, but slower, although probably runs pretty well on GPUs. I would use:
rlist = 1.4 (is optimized by mdrun)
coulombtype = PME
rcoulomb = 1.4
vdw-type = Cut-off
rvdw = 1.4
fourierspacing = 0.18
dispcorr = no

1 Like