[gmx-users] Fwd: Implicit solvent simulations using GROMACS-2016

Sai Pooja saipooja at gmail.com
Wed Apr 4 23:27:10 CEST 2018


I am trying to run an implicit solvent (water) simulation for a protein. I
am able to successfully run one with gromacs 4.5 with charmm forcefield and
infinite cutoffs with the sd integrator. When I follow the same steps and
try to do the same with the newer version i.e. gromacs-2016.1 (same setup
except cutoff-scheme = group), I am unable to equilibrate the system at
constant temperature. I have tried timesteps from 0.002 to 0.0005.
Temperature explodes within the first 100 steps. I am at a loss for
reasons. Is it still possible to simulate implicit solvent systems with

Forcefield: CHARMM (with cmap)
Here are the mdp file parameters:
define                  = -DPOSRES

integrator               = sd
dt                       = 0.001
nsteps                   = 1000

cutoff_scheme = group
vdwtype            = cutoff
ns_type              =  simple
nstlist                 = 0
rlist                     = 0
rgbradii                  = 0
rcoulomb          = 0
rvdw                  = 0
coulombtype    = cutoff
pbc                      = no
comm-mode      = angular
comm-grps        = Protein
implicit_solvent = GBSA
gb_algorithm      = OBC
sa_algorithm = ace_approximation
sa_surface_tension      = 2.25936

nstxout                  = 1 ; ast frame (coordinates)
nstvout                  = 500   ;  (velocities)
nstfout                  = 500   ; No output, except for last frame (forces)
nstlog                   = 1  ; Write every step to the log
nstenergy                = 500  ; Write energies at every step
xtc_grps                 = Protein
nstxtcout                = 1   ; Do not write a compressed trajectory
energygrps               = Protein  ; Write energy information separately
for these groups
; Temperature coupling
tcoupl                   = no
tc-grps                  = Protein
tau_t                    = 2         ; 1/91 2010 gromacs charm paper?does
not work
ref_t                    = 300
gen_vel                  = yes
gen_temp                 = 300

; Pressure coupling
pcoupl                   = no
tau_p                    = 1
ref_p                    = 1
compressibility          = 4.5e-5
refcoord_scaling         = com

constraints              = all-bonds

gmx_mpi mdrun -deffnm proteinsim_ -multi 24

The simulations start but the temperature rises very quickly and then the
simulation explodes. I have explored various options like trying different
temperatures, changing tau_t to slow down and/or speed up temperature
equilibration and tried different timesteps but without much success.

If anyone has a benchmark for running implicit solvent simulations with the
new gromacs (post 4.5), I would be very grateful.

I also see these comments in the logfile:
Using SIMD all-vs-all kernels.

This is simulation 0 out of 24 running as a composite GROMACS
multi-simulation job. Setup for this simulation:

Using 1 MPI process

Your simulation settings would have triggered the efficient all-vs-all
kernels in GROMACS 4.5, but these have not been implemented in GROMACS
4.6 and 5.x. If performance is important, please use GROMACS 4.5.7
or try cutoff-scheme = Verlet.
Does this mean that the older version is recommended for implicit solvent

"Tiger got to hunt, bird got to fly; Man got to sit and wonder, 'Why, why,
why?' Tiger got to sleep, bird got to land; Man got to tell himself he
understand." --KV

"Tiger got to hunt, bird got to fly; Man got to sit and wonder, 'Why, why,
why?' Tiger got to sleep, bird got to land; Man got to tell himself he
understand." --KV

More information about the gromacs.org_gmx-users mailing list