[gmx-users] editing topology to have extra lennard-jones interactions

[Amanda Parker]

Hello GMX mailing list,

I have a topology file that contains most of the force field information I
need (it does not call any external itp or atp files), and I'd simply like
to modify it to have some extra Lennard-Jones interactions with parameters
of my choosing.

The topology originally has the following sections: [ defaults ], [
atomtypes ], [ moleculetype ],  [ atoms ], [ pairs ], [ bonds ], [
exclusions ], [ angles ], [ dihedrals ], [ system ] and [ molecules ].
Using this topology file, I can run grompp and mdrun successfully (no
warnings or errors).

What I can't figure out is

1. How is Gromacs calculating the non-bonded (more specifically, LJ)
parameters for the simulation? Is it using the parameters in [ atomtypes ]
or in [ pairs ]? Does one override the other?

2. If I want to add my own LJ pairs and parameters, where would I add them?
Should I add them to [ pairs ]? I know from the manual (4.6.5):

"Extra Lennard-Jones and electrostatic interactions between pairs of atoms
in a molecule can be added in the [ pairs ] section of a molecule
definition. The parameters for these interactions can be set independently
from the non-bonded interaction parameters."

If I do this, do I need a [ pairtypes ] section? (The original topology
worked fine and didn't need one.)

Alternatively, should I label those pairs of atoms with new atom types and
define their parameters in [ atomtypes ]? Create a new section called [
nonbond_params ] and define the pairs and parameters there?

I have read the manual and mailing list, but most of the advice,
understandably, assumes one is using a built-in force field (like CHARMM,
etc.). You'd think my case would be simpler, but I can't find a straight
answer.

I also know I haven't explained *why* I want to do this, or what I've
already tried, but I didn't want to convolute things. If any of that is
relevant, I can elaborate.

​Sincerely,​
Parker