[gmx-users] Designing a topology file: bonded/non-bonded mess

[Frankie Montenegro]

Hi all,

I am hoping someone could give me advice regarding a topology file design.

I am trying to implement a forcefield for an organic crystal with 30
atoms at each site (therefore, 30 atom basis). Sites are connected by
an aromatic ring.

The forcefield I am trying to implement does not treat ring as a
single unit; rather, carbons are a part of the 3- and 4- body
interactions with other atoms from the site, and two of them have
harmonic term between themselves.

Initially, I tried to set each (30-atom) site as one "moleculetype".
So, half of the aromatic ring would be in one "moleculetype", and half
of it in the neighbor. But then I have to set up BONDED interactions
between atoms from DIFFERENT molecules. And not just one, three of
them: torsion, angle term and multiple harmonic terms.

I was hoping someone could see a solution here. I tried regrouping ring
 carbons to different molecules (sacrifising symmetry) but I always
end up with a bonded interaction between "nonbonded" atoms.

Any suggestion is appreciated.
F.