[gmx-users] Explicitly defined nonbonded LJ pair terms?

[Michael Shirts]

David, Paul, all-

Thanks for the responses -- I've been away from this problem and am
now coming back:

Original question:

> So, I've been reading around, and am trying to figure out if it is
> currently possible in Gromacs to explicitly define separate ij terms,
> i.e., that override the combination rule for specific atom types.  In
> other words, say I have an alkane in water -- I want to vary the C-O
> interaction independent of the C-C and the O-O interaction.  Is there
> a way to do this currently?

Sample answer:

> As long as it's the same potential (i.e. LJ or Buckingham) you can
> specify parameters in the [ nonbonded_params ] section (check spelling)

Follow up questions:

1. Does this work between atoms that are on _different_ molecules
(i.e., alkane in water).

2. When I try this  (inserting it just before the molecular
definitions, and after all the type definitions, as so):

[ nonbond_params ]
;   i       j    func     sig                              eps 
 OW CT    1       0.332794606326    0.433634814359
 OW HT    1       0.281262777488    0.2923565322

I get the grompp error: Fatal error: Atomtype 'OW' not found!

I'm unable to determine why this happen, as the atomtypes are defined
earlier in the file.

3. Apparently, these interactions are set up not to be perturbable by
free energy.  Is this currently correct?  Any suggestion how I might
want to compute using such a perturbative path (other than by writing
the routines to do this)?    For methane, I can just change the C and
H parameters -- for other molecules, this will be more difficult.

Thanks!
Michael Shirts
Columbia University