[gmx-users] Re: A strange phenomenon in the new version opls-aa force field

[Erik Lindahl]

> I noted that you have changed the cetral atoms(N or C) from the 
> position 1 to the position 3.
> I think the atom order of the old version is correct. Becuase in the 
> opls-aa force field, we use
> proper dihedrals [1+cos(2*x+180)] to model improper dihedrals. The 
> improper torsion angle should be
> defined as the angle between planes(C, CA, +N) and (CA,+N,O). But inn 
> the new version *.rtp file, the
> impropers section define the angle between planes(CA,+N,C) and 
> (+N,C,O).
>
> Why you changed the atom order for improper torsion? Could you give me 
> detail explanation?
>

Because that's the way they are defined in OPLS & Amber - there really 
isn't any right/wrong choice - both of them will work :-)

To the first order the potentials will be the same, but in second order 
they differ slightly depending on the order (which changes the 
definition of the planes). All the improper potentials are only 
order-of-magnitude guesses, though, so it doesn't matter...

In the long run I want to have them identical to Amber/OPLS to 
facilitate comparisons, but you can change it back without risking any 
results.

Cheers,

Erik