[gmx-users] NVT simulation and mdp file

[Justin A. Lemkul]

teklebrh at ualberta.ca wrote:
> Dear Gromacs Users,
> 
> I have encountered the following issues while I was running my MD 
> simulation. Can anybody comment on what the meaning of these notes are. 
> Is there anything I could do to avoid them.
> 
> NOTE 2 [file PAP.top, line unknown]:
> 
>   The largest charge group contains 12 atoms.
> 
>   Since atoms only see each other when the centers of geometry of the 
> charge
> 
>   groups they belong to are within the cut-off distance, too large charge
> 
>   groups can lead to serious cut-off artifacts.
> 
>   For efficiency and accuracy, charge group should consist of a few atoms.
> 
>   For all-atom force fields use: CH3, CH2, CH, NH2, NH, OH, CO2, CO, etc.
> 
> My SOLVENT IS TOLUENE --- the PRODRG gave me a topology file with only 
> one group charge only.
> 

That's almost certainly wrong.  See, for instance, the PHE side chain in the 
relevant .rtp entry for a more reasonable charge group setup.  If you're using 
PRODRG defaults, then the charges are probably unsatisfactory, as well.

The rationale for the charge group size is summed up here:

http://lists.gromacs.org/pipermail/gmx-users/2008-November/038153.html

> NOTE 1 [file nvt.mdp, line unknown]:
> 
>   The Berendsen thermostat does not generate the correct kinetic energy
> 
>   distribution. You might want to consider using the V-rescale thermostat.
> 
> 

See the literature about this one, as well as the numerous list archive 
discussions.  For initial equilibration, a weak coupling scheme is probably 
fine, but you can also use V-rescale.  Also of interest:

http://www.gromacs.org/Documentation/Terminology/Thermostats

> NOTE 3 [file aminoacids.dat, line 1]:
> 
>   The optimal PME mesh load for parallel simulations is below 0.5
> 
>   and for highly parallel simulations between 0.25 and 0.33,
> 
>   for higher performance, increase the cut-off and the PME grid spacing
> 

This all depends on the size of your system, how much of the work is distributed 
between the real-space Coulombic interaction and PME.

> 
> In addition to the above notes I have also some questions about  the NVT 
> and NPT simulation.
> 
> 1)I am using toluene as a solvent to simulate my polymer, do I need to 
> use the compressibility of toluene which is  9.2e-5 or the default 
> value  4.5e-5 1/bar.

Well, 4.5e-5 corresponds to water, which you aren't using...

For NVT, this won't matter since the box is fixed, but for NPT, the 
compressibility will affect the response of your system to pressure.  The 
differences may be minimal, but if you know the right value, why accept a wrong one?

> 2)What about the dielectric constant (the dielectric constant for 
> toluene is 2-2.4), but the default value is 80 ( I assume this is for 
> water- am I right).

Yes, the default again assumes water as the solvent.

> 3)Is  always rvdw = 1.4 nm for GROMOS96. As a result I have to increase 
> my box size of the solute at the beginning to a min of 2*1.4 =2.8 ( min 
> image convection). Is this the right way to do!

At an absolute minimum.  Keep in mind that the box vectors will fluctuate under 
NPT, so if the box decreases even a little bit below 2.8, you will be violating 
the minimum image convention.

> 4)I run an NVT simulation to equilibrate my system  for 100 ps. When I 
> checked my simulation at the end (successfully completed) I noticed that 
> the shape of my simulation box looks CIRCULAR! some how the rectangular 
> shape looks distorted. What does this tell! Do you guys think something 
> is wrong in my simulation.

This could be some visualization artifact, or the components of your system have 
condensed within the box.  Without actually seeing it, it's hard to tell.  If 
you post an image online (Photobucket, etc) then we might get a better sense of 
what's going on.

> 5)I included the polar and aromatic hydrogens in my simulation (  
> ffG43a1.itp – GROMOS96.1 in PRODRG). Does these hydrogen influence my 
> result as the force field is a united atom force field. Or How can I get 
> rid of them if I want. With or without the aromatic hydrogen gave good 
> results ( besides lower computational cost). Does Gromos96 model 
> correctly aromatic-Aromatic interaction.
> 

Well, "correct" is a relative term for all force fields, but you need to follow 
the prescribed setup of the force field itself, otherwise you can throw it all 
away.  If you lump the hydrogens into the ring carbons and have an uncharged 
ring, the result will be different than if you have the hydrogens there with a 
small charge on each C and H.  Again, refer to the force field .rtp file for 
examples.  You can also create a better toluene topology by renaming the residue 
in your coordinate file PHE and trick pdb2gmx:

pdb2gmx -f toluene.pdb -missing

Then change the mass of the CH2 group (which pdb2gmx thinks is a CB for PHE) to 
reflect a CH3 group.  Make an .itp file out of the resulting .top by removing 
the unnecessary #includes, [system], and [molecule] directives.  Then you don't 
have to worry about messing with PRODRG.  I should note, as well, that this is 
about the only appropriate use of -missing I can think of at the moment (just 
for clarity in the archives; I usually warn against using -missing).

-Justin

-- 
========================================

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

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