Re: [AMBER] reproducing a simulation

From: Jason Swails <jason.swails.gmail.com>
Date: Sat, 23 Jul 2011 12:34:13 -0400

You will most likely need a longer simulation. Depending on the size of the
system, you may need a longer simulation than even 5 ns. (And you also need
more than just RMSD to determine simulation similarity).

Did you also use the same random seed? I would suggest setting ntpr=1 and
seeing if the beginning of the simulation is identical (if it's not, not all
of the settings are identical).

HTH,
Jason

On Sat, Jul 23, 2011 at 8:45 AM, Xiao Chen <chen.xiao.po.gmail.com> wrote:

> Thanks Jason!
>
> I reran the simulation using identical parameters for 200 ps and collected
> 20 frames over this time.
> Then I compared the overall r.m.s.d using ptraj of these 20 frames with the
> 20 frames of the original simulation (of 5ns). The r.m.s.d values shown in
> the graph are for
> 1) first 20 frames of the original simulation and
> 2) 20 frames for the rerun of the simulation
>
> with the same reference structure.
>
> So can I say that the results from the original have been reproduced or do
> I
> need to do a longer run of 5ns to
> see if the results are identical/similar.
>
> Thanks much!
>
> Po
>
>
>
> On Fri, Jul 22, 2011 at 8:03 PM, Jason Swails <jason.swails.gmail.com
> >wrote:
>
> > On Fri, Jul 22, 2011 at 7:34 AM, Xiao Chen <chen.xiao.po.gmail.com>
> wrote:
> >
> > > Dear All,
> > >
> > > I ran a molecular dynamics simulation of the exact same ligand-protein
> > > complex, using the exact same parameters and input files ,twice. After
> > > generating 20 frames (each) ,I calculated the rmsd (using ptraj) for
> each
> > > and got the attached numbers.
> > >
> >
> > 20 frames is ambiguous. Does that mean you ran the simulation for 20
> steps
> > and collected a snapshot each step? Was the total simulation 20 ps? The
> > shorter the simulation, the less likely the simulations are identical if
> > they've begun to diverge.
> >
> >
> > > Can we say that the two simulations are identical, and that it has been
> > > reproduced, based on these attached numbers?
> > >
> >
> > Your RMSD values are not identical -- one is systematically smaller than
> > the
> > other. Furthermore, RMSD is not a sufficient metric to determine if 2
> > simulations are identical (it can be enough, however, to declare that
> they
> > are not). Consider what RMSD actually tells you: it tells you the
> average
> > deviation of atomic positions. It says nothing about which direction the
> > deviations occur in. Therefore, 2 structures with a 5 Angstrom RMSD from
> a
> > common reference can be anywhere between 0 and 10 Angstroms deviation
> away
> > from each other (over-simplified, of course, but you get the idea).
> >
> > However, were you even using the same reference in these 2 cases? (You
> > should be).
> >
> > HTH,
> > Jason
> >
> >
> > > Thanks,
> > >
> > > Po
> > >
> > > _______________________________________________
> > > AMBER mailing list
> > > AMBER.ambermd.org
> > > http://lists.ambermd.org/mailman/listinfo/amber
> > >
> > >
> >
> >
> > --
> > Jason M. Swails
> > Quantum Theory Project,
> > University of Florida
> > Ph.D. Candidate
> > 352-392-4032
> > _______________________________________________
> > 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
>



-- 
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Jul 23 2011 - 10:00:03 PDT
Custom Search