[gmx-users] Improper Dihedrals

[Erik Lindahl]

Dallas Warren wrote:

> David,
>
>> based on atomtype.
>> Check the type your atoms have in the [ atoms ] section
>
>
> Actually, that isn't quite what I am asking.
>
> In the bon.itp file for the force field, there are two constants 
> defined for i and l  These are some how put together to form an 
> improper dihedral for i, j, k, l.  In the case of an amide bond (with 
> ffgmx), you have atom types CH2, C, OM and N.  Therefore, which 
> improper dihedrals have to be defined in ffgmxbon.itp to form the 
> final description of it?  CH2 - N, CH2 - OM and N - OM?

Hi Dallas,

First, if you check ffoplsaabon.itp you'll see that it is actually 
possible to define proper & improper dihedrals with four atomtypes now 
(not only two). I'll add it to the documentation for 3.2, but we wanted 
to see that it worked first.

Second, I find it easier to think in terms of the planes defined by atom 
triplets. Both for proper and improper dihedrals there are four atoms 
'i','j','k','l'. If you use the mathing rules based on two atomtypes 
only, it will be j & k for proper dihedrals and i & l for impropers.

The actual dihedral angle is then defined as the angle between the two 
planes defined by i-j-k and j-k-l.

Personally I always make mistakes when I try to think in ways of 
'central atoms', since there are too many permutations... :-)

Cheers,

Erik