I switched this to the development list ...
I think it's a good point here that we need reference enemies (and forces)
for systems with enough variety that it hits all terms in the database.
Something to discuss at the amber meeting maybe...
On Sep 16, 2014 5:43 PM, "Jason Swails" <jason.swails.gmail.com> wrote:
>
> On Sep 16, 2014, at 4:11 PM, David A Case <case.biomaps.rutgers.edu>
> wrote:
>
> >> On Sun, Sep 14, 2014 at 7:46 PM, David A Case <case.biomaps.rutgers.edu
> >
> >> wrote:
> >>
> >>> It can be tricky to get everything correct. The 2009 Best/Hummer
> >>> modifications involve only corrections to the psi angle, but you have
> to
> >>> make sure that they are not applied to glycine or proline. I'm not
> sure
> >>> that this can be done inside tleap without making some new atom types
> that
> >>> are specific to GLY and PRO. This might be a place where a ParmEd
> script
> >>> could be of real use.
> >
> > I figured if I mentioned ParmEd, that Jason would have had a solution by
> > now. :-) Of course, as others have mentioned, the problem is not so
> much
> > doing this (you could even hand-edit your ff99SB-ildn prmtop file to
> make it
> > ff99SB*-ildn), as in making sure it agreed with GROMACS, or whatever
> other
> > program (desmond?) provides the "reference implementation" for
> ff99SB*-ildn.
>
> Haha. Well it _can_ be done with ParmEd. It has all of the functionality
> required to adjust parameters — setBond, setAngle, addDihedral, and
> deleteDihedral. And for certain classes of modifications you could
> implement the force field with a small number of ParmEd commands. If
> you’re modifying backbone dihedrals this becomes significantly harder to
> automate (because the selection masks will not be constant).
>
> If all of the modifications are intra-residue (like side chain torsions),
> then ParmEd may be the easiest solution. If not, then any ParmEd-based
> solution will have the same challenges as the “right” way to do it through
> tleap and frcmods.
>
> > Force field interoperability has made significant strides recently,
> mostly by
> > programs that manipulate "low-level" descriptions like prmtop, PSF or
> tinker
> > analyze files. There are lots of Google hits (including at the Amber web
> > page) for programs to convert Amber topologies to Gromacs, but I'm not
> > personally aware of work in the other direction. (Pipe up if you know
> how to
> > do this). That sort of tool would enable users to build the system in
> > Gromacs, then run it in Amber.
>
> This almost made it into ParmEd this summer, actually… my summer just
> ended 2 weeks too early :). To validate against GROMACS, you can always
> implement ff99SB*-ildn in Amber, convert the prmtop to GROMACS, and make
> sure they give exactly the same energies as the “reference” implementation
> for a wide array of systems.
>
> All the best,
> Jason
>
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue Sep 16 2014 - 20:30:02 PDT
