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

[Frankie Montenegro]

Wow, I didn't realize you guys in Europe work 24/7 too. On Sunday morning
nonetheless :-)

On 10/14/07, David van der Spoel <spoel at xray.bmc.uu.se> wrote:

> question: are there chemical bonds between "molecules" (which you seem
> to imply by stating the sites are connected by an aromatic ring). If
> there are then your whole system should be considered as a single
> molecule. Have you tried x2top on this? (gmx 3.3.2 only).
>

Yes they are. x2top wouldn't work, most atoms are redefined (two types
of O with
different charges, two types of C, etc. All are custom-fitted with
Gaussians and GULP) I don't think treating the whole system as one
molecule could work. A single "cell" of the crystall would have over
250 atoms, and I need 2x2 structure minimum, probably even bigger. But
that might be last resort (after all , aren't some proteins over
couple of thousand atoms?)

BTW, what is the slope of GMX_performance vs. molecule_size curve ?

There is one more regrouping that I am trying right now (namely having two
moleculetypes: "cube-edge" and "cube-node", adding some  table "LJ" in between).
 If this doesn't go, I am out of ideas.  Thanks anyway.