Good point. I encountered the same problem when I optimized amino-acid-like
small molecules using Gaussian. The optimizer in the recent Gaussian
versions can migrate proton from one heavy atom to another to neutralize
the molecule. The two atoms may be separated several bonds (the longest
distance is 7 bonds)! My solution was to use PCM to do optimization and
then did a single point calculation to calculate ESP. With PCM, the charge
centers are well kept.
Best
Junmei
On Sat, Oct 28, 2017 at 8:16 AM, David A Case <david.case.rutgers.edu>
wrote:
> On Sat, Oct 28, 2017, Andreas Tosstorff wrote:
>
> > Thanks a lot for the advice! The structure I have is that of the small
> > molecule bound to a protein. If I submitted this structure, without
> > further geometry optimization, wouldn’t that overstabilize the bound
> > ligand conformation?
>
> In principle, MD results should become independent of the starting
> configuration as sampling becomes more complete.
>
> >From the point of view of developing charges, many people prefer to fit
> several conformations at once, to get kind of an "average" charge set that
> works pretty well for many conformations. The ways in which those
> conformations are generated is outside the scope of antechamber/resp.
>
> You might try adding a cosmo solvation model to the Gaussian run, then
> optimizing. The will probably stabilize the zwitterionic form. Or,
> minimize
> the molecule in a force field, since that is what you will be using for the
> binding calculation anyway. (This is iterative: get charges for
> the conformation you have, minimize in ff, re-determine charges from this
> minimized conformation, see if they have changed very much.)
>
> ....dac
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Oct 28 2017 - 06:30:03 PDT
