[gmx-users] 3-site methanol model (Re: C. Freudenberger)

[Christoph Freudenberger]

Hi Nuno,

Nuno R. L. Ferreira wrote:
> Yes, you may say I'm lucky. Choose one, you may say that. Actually, I've saw
> your box in the molecules directory in GROMACS contributions, in April. When
> I was talking about the 3-site model, I was refering to your methanol box
> contribution.
> 
> Already runned a 1 ns MD with your equilibrated methanol box, and the Dt is
> OK, 1.6 10-5 cm2 s-1, just like the one presented by WFG in his 2000 JCP
> article.
> The 6-site methanol model that I talked about in the last e-mail posted in
> this gmxuser's list , is from charmm forcefield. But this molecule is
> categorized in the forcefield as a model molecule, so its still on a
> experimental basis, I presume. 
You mean the model is still in development status?

> As you might know ,the academic version of
> CHARMM has no support, so I'm alone to talk about their model. Have to do a
> net search about using methanol within CHARMM.
One more reason to use GMX. And one more cheer to the GMX-team for their
patience with us ;-)

> 
> My questions are :
> 
> 1 ) I want to do a MD simulation of a peptide in methanol. Right know, I'm
> interested in the diffusion coefficient of the protein and also on the Dt of
> the solvent, not just the bulk one, but to see if I can descriminate a kind
> of hydration shells around the solute. If your model is good enought to
First of all it's not 'my' model, I just wrote a gromacs topology for the
WFG model (denoted B3 in that JCP paper).
> mimic the bulk methanol experimental Dt, think might give me other good Dt's
> , concerning the solvent, in comparison to the 6-site model, that, as I said
> last e-mail, was about 1.8 10-5 cm2 s-1. Is this "true" ?
AFIAR g_msd gives you an error estimation for Dt. Compare the 3- and the
6-site model within those error bars. I suppose you have to consider both
models equivalent concerning the diffusion.

> 2) When performing methanol Dt calculations around the peptide, what is the
> best way to do this ? I mean, should I fixe the peptide all along the MD, or
> let it translate?

Personally I would avoid any artifical influences on the system if there is
no actual need to do so. Hence I would not use any contraints or freeze groups.
But being an organic Chemist my molecules are way smaller.
BTW... I'm just wondering what happens in a protein box simulated with comm...
I figure as a large part of the systems mass belongs to the protein, the
protein might propably not diffuse at all... Is that correct?

> 3) Took a look at your methanol .itp. Don't understand how you translated
> the Lenard-Jones parameters from WFG JCP 2000 article to your topology file.
> It's just a curiosity. If I have some more doubts about your meoh box
> implementation, I'll contact you on this list
In the paper the the LJ's are given in terms of sqrt(C12) or sqrt(C6) respectily.
I suppose that's due to the kind of combination rule used.
So for the .itp I just squared the values, e.g.:
C12(O)=(1.525e-03)^2=2.325625e-6 [kJmol/nm^12]

-- 
Christoph Freudenberger
Abt. Organische Chemie I AK Siehl - Uni Ulm -Tel: ++49-731-502-2785