I agree with Jason, but also want to add my concern with watching the time
dependence of the REMD in this way. The fact that it originally "matched"
the experiment but later did not match should indicate that no matter which
force field you end up using, you need a way to reasonably estimate the
precision. A variety of "bad" force fields may result in the simulation
remaining in the native state for different lengths of time (as will
different REMD runs). None of those differences are directly indicative of
relative stability. You probably are looking for a measure of thermodynamic
stability under those conditions, which will need precision estimates (best
done from multiple runs initiated with different conformations). A short
REMD run that stays in the initial structure (whether it is folded or not)
won't tell you much about thermodynamic properties. Stopping the simulation
at a point where it matches the experiment isn't usually good science, in
my opinion.
On Thu, Mar 27, 2014 at 11:14 AM, Jason Swails <jason.swails.gmail.com>wrote:
> On Thu, 2014-03-27 at 20:33 +0530, Gargi Borgohai wrote:
> > Dear AMBER Users,
> >
> > I wanted to carry out an REMD simulation of a 20-residue mini-protein in
> 6M
> > urea solution. I have used amberff94 force field.
>
> This force field is ~20 years old with well-known and well-documented
> deficiencies. There is no reason to continue using it (it was a great
> contribution at the time and served as an excellent foundation for later
> efforts, but ff94 is a horrible choice today).
>
> > As reported in
> > literature, the native state should retain 60% of folded state (in 6M
> urea
> > solution) at 280K with REMD simulation. In my study the results matches
> > till 20ns of simulation. But beyond 20ns the folded fraction of the
> peptide
> > abruptly decreased to 33%.
> >
> > Should I stop the simulation at 20 ns and start analysing the results?
>
> No. Use a better force field. There's a chapter in the Amber manual
> regarding force fields. The currently recommended choice is ff12SB.
>
> Good luck,
> Jason
>
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
>
>
> _______________________________________________
> 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 Thu Mar 27 2014 - 09:30:02 PDT
