The scatter would be because the different force fields give different
relative energies for the various conformations - that's part of being a
different force field. The shift is because the zero in MM energy is
arbitrary and not included as a target in training.
On Sun, Nov 12, 2023, 4:12 PM Albert Gauss via AMBER <amber.ambermd.org>
wrote:
> Hi Carlos,
>
> thank you for your quick response, which I found very useful. I plan to
> follow your advice and will share the results.
>
>
> I'd like to ask one follow-up question.
>
> Could you please briefly outline the reasoning for the observed energy
> shifts and scattering?
>
> Thanks a lot for your help.
>
>
> Best,
>
> Albert
>
>
> On 11/10/23 17:04, Carlos Simmerling wrote:
> > Hi Albert, see my comments below
> >
> >
> > On Fri, Nov 10, 2023 at 6:31 AM Albert Gauss via AMBER
> > <amber.ambermd.org> wrote:
> >
> > Hi all,
> >
> > I've been using the amber99SB-ILDN FF in Gromacs for a while now. I
> > hadn't really considered switching to a newer version until
> recently.
> >
> >
> > I think that would be good, this is a very old FF now and isn't
> > necessarily bad, but you should be using more recent/better models.
> >
> >
> > It's worth noting that Gromacs doesn't officially ship with force
> > field
> > versions higher than 99SB, but I'm interested in exploring those
> > options. Lately, I've been building my systems in Leap and then
> > converting them via Parmed for the newer versions. However, I'm
> > unsure
> > about the reliability of these topologies. To gain a better
> > understanding, I decided to calculate different energies for the same
> > trajectory using all three force fields and compare the results. My
> > primary concern is the shift in potential energies between these
> > force
> > fields.
> >
> >
> > yes they should be shifted, of course, and perhaps with some scatter
> > as well.
> > It might be good to do the same analysis in Amber to get "reference"
> > energies for the implementation.
> >
> > Additionally, I observed that the coulomb energies are
> > consistent between 99SB and 14SB but differ from SB19.
> >
> >
> > This should not happen, there are no changes to the partial charges in
> > ff19SB.
> > I suspect an implementation problem, since that has been known to
> > happen in the past.
> > For example, see the correction published by Best et al due to faulty
> > ff99SB implementation in Gromacs
> > https://www.cell.com/biophysj/fulltext/S0006-3495(08)78586-2
> > I would recommend evaluating the energies in Amber to get "correct"
> > ff19SB energies.
> >
> >
> > Overall, I would greatly appreciate any feedback on the
> > differences I've
> > mentioned and the overall reliability of the conversion procedure
> > from
> > Amber to Gromacs via Parmed.
> >
> > Thank you in advance for your insights and assistance.
> >
> >
> > Best
> >
> > Albert
> > _______________________________________________
> > 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
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun Nov 12 2023 - 14:30:04 PST
