[gmx-users] opls parameters

Erik Lindahl lindahl at cbr.su.se
Sun Dec 3 12:25:21 CET 2006

Hi Mark,

On Dec 3, 2006, at 7:20 AM, Mark Abraham wrote:

> So evidently opls_005 is a nitrogen atom... so what's with the "C2   
> 6"?
> Further, there is no bonded interaction in ffoplsaa.rtp for "C2", but
> there is C_2... So I really don't understand what is going on.

The "6" doesn't belong to the bonded_type; it's the atomic number of  
the atom type (or heavy atom, in the case of united atoms) and used  
for QM/MM stuff. We added it to most force fields with a script, but  
apparently forgot to update the comment.

The other problem is that this data originally comes from Bill  
Jorgensen's distribution, and it's not entirely complete. For a  
start, the first 65 atoms are for the united atom force field, and  
most of these do not have complete bond information (since the  
original parameter files were the all-atom opls version).

I still decided to keep atoms 1-65 for two reasons: First, I couldn't  
see any good reason to remove them when they only used 65 lines, and  
second there are a couple of useful ones without bonds (united-atom  
CH4, ions, etc.).

However, and this is the good news, OPLS bonds and angles are  
essentially amber, and the united-atom ones likely correspond to  
amber. The only exception here is for the newer torsions, where we  
have also had to introduce some extended types (anything X_2 is  
derived from X) where Bill introduced "exceptions" for particular  
residues or atoms.

If you need to automate conversion, my first recommendation would be  
to cut atoms 1-65 and try again!



More information about the gromacs.org_gmx-users mailing list