Hi all,
I am having great difficulty taking my Gromacs MARTINI coarse grain model of a protein to a dt of at least 0.025 fs. I have modelling my protein using the Martinize of an atomistic structure, and I have replaced most of the equilibration angles/distances using those derived from the atomistic simulation (although, they weren't that far off with to begin).
Starting with 0.0005 ps, I ran my system for 300000 steps, then using the same number of steps I moved it through to 0.001, 0.0015 and finally 0.002 ps. Any attempt at taking the dt beyond 0.002 results in my themostat going wildly out of control (and pressure when I've applied a pressure coupling) before crashing out with LINCS warnings.
I've put the NVT settings below (although note, lincs continuation and genvelocities change depending on what stage in equilibration you are in).
Any thoughts on how I can resolve this issue?
title = Martini
integrator = md
dt = 0.025
nsteps = 600
nstcomm = 100
nstxout = 100
nstvout = 100
nstfout = 0
nstlog = 1
nstenergy = 1
nstxout-compressed = 0
compressed-x-precision = 0
;compressed-x-grps =
energygrps = collagen solvent
cutoff-scheme = Verlet
nstlist = 10
ns_type = grid
pbc = xyz
verlet-buffer-tolerance = 0.005
coulombtype = PME ;reaction-field
rcoulomb = 1.1
fourierspacing = 0.16 ;0.2 ;0.12
epsilon_r = 15 ; 2.5 (with polarizable water)
epsilon_rf = 0
vdw_type = cutoff
vdw-modifier = Potential-shift-verlet
rvdw = 1.1
tcoupl = berendsen ;v-rescale ;berendsen ;v-rescale
tc-grps = collagen solvent
tau_t = 0.5 0.5
ref_t = 310 310
gen_vel = yes
gen_temp = 310
gen_seed = 473529
continuation = no
constraints = none
constraint_algorithm = lincs
lincs-warnangle = 45
lincs-order=18
lincs-iter=4
What happens if you use the standard force field parameters? Do you include explicit waters? For a delta-t of 2 fs, you are using all-atom model. The coarse grained model allows in principle a much larger integration timestep, of at least 10 fs.
The original all-atom model used explicit water molecules, as does the CG model. I am not sure what you're trying to tell me from your post.
You haven't provided with much detail of your system and what you did prior to the NVT, but what I can tell from the information you provided are the following.
1) your timestep is too small for coarse-graining. It is usually 0.02 - 0.04, with few exceptions. Like Martini tutorial for polymer suggests the time step should be lower than 0.02 when considering dihedrals. In the mdp file you mentioned 0.025 but above you mentioned 0.0005 ( what is this timestep for really confused about this)
2) Coloumbtype for Martini is shift and not PME ( remove all the parameters dependent on PME scheme)
3) cutoff-scheme is group
4) vdw_type is shift
5) I am not very sure about this but I think tcoupl should be v-rescale
6) Usually, during equilibriation we add some constraints with define=-DPOSRES ( this again I recently realized but I am not very sure on how this works, also maybe try out constraints=all-bonds)
7) Also rcoulomb =1.2 rcoulomb-switch =0.0, rvdw=1.2, rvdw-switch=0.9
The parameters are essential to use martini models as all the topology files reported are found using these parameters and you can find these parameters in Martini 2007 paper on lipids.
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Biomolecules
- Proteins