[gmx-users] gromacs charges

[Erik Lindahl]

quantix1 at gmx.de wrote:
> Dear Gromacs-Users,
> 
> I am completely new to Gromacs. Let's say, I'm still thinking about using
> it. My first question is, how charges for non-standard amino acids or other
> molecules are usually derived for the force field in Gromacs. Is there a
> procedure which is considered to be standard and gives the most 
> consistent results (like the HF/6-31G -> RESP for AMBER) ?
> This should be enough before the weekend. Thanks in advance,
> 
> Uwe
> 

When it comes to building new molecules, nothing beats the Jorgense 
OPLS-AA/L forcefield (yes, it's available in Gromacs even if the 
webpages don't mention it). The nice thing about OPLS is that all groups
are designed and parameterized as "building blocks", so you can attach a
new functional group to e.g. a protein, and you will only have to update
the charge of one or two atoms. Check the file share/top/ffoplsaa.atp 
for comments on all the different atom types.

OPLS charges are derived with CHELPG, and for proteins all the backbone
dihedrals have been parameterized not only to reproduce the position of
the minima, but the actual form of the energy landscape.

The AMBER forcefield is on our todo-list, but we've never felt really 
motivated since OPLS/AA is better in most cases.

Cheers,

Erik