[gmx-users] System heating up during MD run? (Implicit solvent)

[Matt Larson]

I've been having problems getting implicit solvent systems (which are
probably fairly experiment still in gromacs) to work correctly.  I've
been modelling a protein of about 11000 atoms with hydrogens in a 2 ns
simulation.  By the end of the simulation, the temperature has risen
from 300 K to 496 K - and the protein unfolds.  It was supposed to
maintain temp at 300 K.

I have an a non-zero total charge of -6.999999.  With explicit
solvent, I would normally add the appropriate matching ions, but with
implicit solvent what should you do?   Could the non-zero total charge
result in heating?  Or is the thermostat not working well enough
(should I change tau-t or ref-t?)

Here is my md.mdp file:


constraints         =  all-bonds ; trying..
integrator          =  md
dt                  =  0.002    ; ps !
nsteps              =  1000000
nstlist             =  10
ns_type             =  grid
rlist               =  1.0
coulombtype         =  cut-off
fourierspacing      =  0.16
vdwtype             =  cut-off
rcoulomb            =  1.0
rvdw                =  1.0      ; important for neighbor searching
pbc                 =  no       ; no periodic boundary conditions
epsilon_rf          =  0
rgbradii            =  1.0      ; must equal rlist, rcoulomb, rvdw
comm_mode           =  angular
optimize_fft        = yes

implicit_solvent    = GBSA
gb_algorithm        = OBC
gb_epsilon_solvent  = 80
sa_surface_tension  = 2.25936

nstcomm         = 10
nstxout         = 1000
nstxtcout       = 1000
nstvout         = 0
nstfout         = 0

tcoupl              = andersen
tc-grps             = system
tau-t               = 0.1
ref-t               = 300

gen_vel             = yes
gen_temp            = 300
gen_seed            = -1

-----