[gmx-users] Implicit solvent MD is not fast and not accurate.

[Justin Lemkul]

On Wed, Mar 27, 2013 at 9:27 AM, xiao <helitrope at 126.com> wrote:

> Dear Gromacs users:
> I did a protein MD using implicit solvent and Amber 99SB force filed.
> However, i found that the implicit solvent is not faster than explicit
> solvent, and what is worse is that it is also not accurate.
> The system is a protein-ligand complex. Firstly, i run a minimization, and
> then i did a production MD. The explicit solvent MD can give nearly same
> strucuture as the crystal structure after 10ns MD. However, there is a
> significant change in the ligand after 1ns MD in implicit solvent.
> My .mdp file is as follows:
> title  = OPLS Lysozyme MD
> ; Run parameters
> integrator = md  ; leap-frog integrator
> nsteps  = 10000000 ; 2 * 500000 = 1000 ps, 1 ns
> dt  = 0.002  ; 2 fs
> ; Output control
> nstxout  = 1000  ; save coordinates every 2 ps
> nstvout  = 1000  ; save velocities every 2 ps
> nstxtcout = 1000  ; xtc compressed trajectory output every 2 ps
> nstenergy = 1000  ; save energies every 2 ps
> nstlog  = 1000  ; update log file every 2 ps
> ; Bond parameters
> continuation = yes  ; Restarting after NPT
> constraint_algorithm = lincs ; holonomic constraints
> constraints = all-bonds ; all bonds (even heavy atom-H bonds) constrained
> lincs_iter = 1  ; accuracy of LINCS
> lincs_order = 4  ; also related to accuracy
> ; Neighborsearching
> ns_type  = grid  ; search neighboring grid cells
> nstlist  = 5  ; 10 fs
> rlist  = 0  ; short-range neighborlist cutoff (in nm)
> rcoulomb = 0  ; short-range electrostatic cutoff (in nm)
> rvdw  = 0  ; short-range van der Waals cutoff (in nm)
> ; Electrostatics
> coulombtype = cut-off ; Particle Mesh Ewald for long-range
> vdwtype         = cut-off
> pme_order = 4  ; cubic interpolation
> fourierspacing = 0.16  ; grid spacing for FFT
> ; Temperature coupling is on
> tcoupl  = V-rescale ; modified Berendsen thermostat
> tc-grps  = system  ; two coupling groups - more accurate
> tau_t  = 0.1    ; time constant, in ps
> ref_t  = 300   ; reference temperature, one for each group, in K
> ;
> ;
> comm-mode           =  angular
> comm-grps           =  system
> ;
> ;
> pcoupl  = no ; Pressure coupling on in NPT
> pbc      = no  ; 3-D PBC
> gen_vel             =  yes
> gen_temp            =  300
> gen_seed            =  -1
> ;
> ;
> implicit_solvent    =  GBSA
> gb_algorithm        =  OBC ; HCT ; OBC
> nstgbradii          =  1
> rgbradii            =  0   ; [nm] Cut-off for the calculation of the
> Born radii. Currently must be equal to rlist
> gb_epsilon_solvent  =  80    ; Dielectric constant for the implicit solvent
> ; gb_saltconc       =  0     ; Salt concentration for implicit solvent
> models, currently not used
> sa_algorithm        =  Ace-approximation
> sa_surface_tension  = -1
>
> Can anyone give me some suggestions?
>

Performance issues are known. There are plans to implement the implicit
solvent code for GPU and perhaps allow for better parallelization, but I
don't know what the status of all that is.  As it stands (and as I have
said before on this list and to the developers privately), the implicit
code is largely unproductive because the performance is terrible.

As for the accuracy assessment, I think you need to provide better evidence
of what you mean. A single simulation is not definitive of anything, and
moreover, some differences between explicit and implicit are likely given
the lack of solvent collisions. The implicit trajectory will probably
sample states that are inaccessible (or at least very rare) in the explicit
trajectory.

-Justin

-- 

========================================

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540)
231-9080http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================