[gmx-users] correction in oplsaa to rb conversion?

[Erik Lindahl]

Hi,

On May 2, 2006, at 9:48 AM, David van der Spoel wrote:

> gil claudio wrote:
>> I just derived the conversion of the dihedral
>> constants from OPLSAA to RB and I get a set of
>> equations different from the ones printed in the
>> manual (3.3, p 56).
>> C1 = V1 / 2 - 3 V3 / 2
>> C3 = 2 V3
>
>> These are the negative of what is printed in the
>> manual.
>
> I think you're right. Question now is what was used in the  
> derivation of the parameters? Erik, can you comment?
>

The equations in the manual are correct, although I did rederive  
everything from scratch during the first port.

I bet both of you are forgetting the conversion between polymer and  
biochemistry convention. That will give you a minus sign for all odd  
powers of cos(phi).

> Also the reference (60) is incorrect, it should be
> Jorgensen, J. Phys. Chem. 90 (1986) 1276
>
> Finally, do also note that in the latest OPLS paper
> Jorgensen & Tirado-Rives J. Comp. Chem. 26 (2005) 1689
> a further term is introduced in the torsion function. This is not  
> yet used in GROMACS AFAIK.

It was present already in the 2001 OPLS-AA/L paper, and we do use it,  
although only for a handful of torsions.

The 2005 paper is only a summary/presentation of MCPRO and BOSS as  
far as I know - no new data for the force field.

Cheers,

Erik