Re: [AMBER] MMGBSA energy calculation problem for RNA

From: Carlos Simmerling <carlos.simmerling.gmail.com>
Date: Wed, 25 Sep 2019 12:06:16 -0400

regarding the difference in energies reported by igb=5 vs 8, igb=8
(GBneck2) included RNA duplexes in both the training and test sets, and
reproduced the solvation energies from PB fairly well. (see
https://pubs.acs.org/doi/full/10.1021/acs.jctc.5b00271 , including the SI
figures for RNA solvation energies). We didn't compare to igb=5 in that
paper, for for the protein GBneck2 model we found that GBneck2 reproduced
absolute solvation energies better than igb=5 (table 3 in
https://pubs.acs.org/doi/full/10.1021/ct3010485).

we optimized these GB models to be fast approximations of PB for use in MD
simulations. If you're postprocessing snapshots and don't need very fast
energies (or any forces), I think you're better off using PB for such
calculations, especially for RNA/DNA. GB is pretty approximate and has
valuable use cases, but I don't think it is the best choice for snapshot
postprocessing unless you have so many structures that PB isn't an option.
carlos

On Wed, Sep 25, 2019 at 11:18 AM Liao <liaojunzhuo.aliyun.com> wrote:

> Thanks there Professor Case,
>
> As I found out that igb=5 and igb=8 made a huge difference, though I’m yet
> to find out why.
> Yes doing it by hand is essential, though I still need to figure out
> sander’s commands’ compatibility between old and new Amber-some of them
> seem to be incompatible, at least from my preliminary try. Still checking
> Amber 19’s manual for the newest by-hand MMGBSA examples.
>
> > On Sep 25, 2019, at 9:54 PM, David Case <david.case.rutgers.edu> wrote:
> >
> >> On Wed, Sep 25, 2019, Liao wrote:
> >>
> >> The issue is that I’ve ran the exact same molecule before with AMBER11,
> >> and the binding energy result was very reasonable, at -71 kcal/mol. The
> >> only significant system setup difference I can think of is that back
> with
> >> AMBER 11 there was no charge neutralization via Tleap with Na+ for the
> >> solvated complex, and now with AMBER 19 22 Na+ were added to neutralize
> >> the system.
> >
> > The way to proceed here is to run MM-PBSA "by hand" so that you know
> > exactly how things like ions are being handled. (See the coments at the
> > beginning of Chap. 32 in the Reference Manual.)
> >
> > The "classic" implicit solvent approach is to consider mobile ions like
> > Na+ as the part of the solvent, that is, to strip then out before doing
> > any analysis. It's not clear (to me) how well alternative approaches
> > that try to treat ions as a part of the RNA may be expected to work.
> > Nor do I know (although others on this list may know) how various
> > automated scripts between Amber11 and Amber19 might have handled this.
> > Hence the recommendation to make sure you understand in detail exactly
> > what the automated scripts are going.
> >
> >>
> >> From the current energy results, the G for the gas phase dsRNA complex
> >> seems to be quite high (compared to the 2 single strands) like as if the
> >> negative charges on RNA are repulsing the 2 strands away.
> >
> > This is exactly what I would expect to be happening. Solvent screening
> > (from both dielectric effects and from mobile counterions) will
> > mostly cancel this electrostatic repulsion.
> >
> > ....good luck....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
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Sep 25 2019 - 09:30:02 PDT
Custom Search