On Jun 17, 2009, at 12:47 PM, David A. Case wrote:
> Hi everyone:
>
> I'm attaching a paper sent me by Edyta Malolepsza, pointing out that
> the
> Amber force fields are not invariant to permutations among
> equivalent atoms
> in a variety of cases. (For example, if you swap OD1 and OD2 in
> asp, you
> might get a slightly different energy.) There is also a proposed fix.
>
> I don't think we can just patch the prep files at this point, since
> that
> would upset backwards compatibility. But I am thinking that we
> should put
> these changes into the next versions of all our force fields, since
> that
> would just represent one improvement (among others) of the new ff's
> over the
> old ones.
>
> There remains the question of how to implement this; ideally, tleap
> and sleap
> would be modified, somehow(?) knowing when to use the new rules and
> when to
> use the old ones. Alternatively, we could post-process the prmtop
> files, as
> suggested in the paper.
It is very easy to make changes to sleap. I think we can do what Ross
suggested for the 14 scale
factor. We add a new section to the frcmod file, which lists all
improper torsions
which should be handled using the new rule. tleap will ignore this
section and treat
all improper torsions using the old rule (so that backward
compatibility is kept), while
sleap will read this section and handle them different.
>
> Comments are welcome, both from developers and from people in the
> Amber FF
> consortium.
>
> As an aside, I'm eager to hear about progress in efforts to release
> updated
> force fields. I have some suggested changes for ff99SB and ff03
> (involving
> LJ 6-12 parameters on His, Lys and Arg sidechains) that I think are
> worth
> testing. But I'm also aware that we don't want to put out force field
> updates in a piecemeal fashion, and I'd be eager to collaborate.
>
> ..thx...dac
>
> <symmetrization.pdf>_______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Mon Jul 06 2009 - 09:56:05 PDT
