[gmx-users] mdrun -rerun results depend on starting .gro file. Why?

[David Mobley]

On 10/24/05, David <spoel at xray.bmc.uu.se> wrote:
>
> On Mon, 2005-10-24 at 15:35 -0700, David Mobley wrote:
> > David,
> >
> > That's what I thought, but the fact is, it IS making a significant
> > difference (several percent). I generate EXACTLY the same tpr file,
> > with the exception of the .gro file and nstlim=1, and I get different
> > energies by several percent. If I use the same .gro file I get the
> > same energies. So clearly there is some interplay going on. I've tried
> > this multiple different times on a number of different systems (in
> > 3.2.1) and the effect is quite reproducible. I am NOT doing something
> > drastic like running it on a different system or anything -- I'm
> > running the SAME system, with everything set the same, except
> > nstlim=1, and a different .gro file for grompp.
>
> You do not say which energies. The kinetic energy might be affected if
> you have no velocities stored (e.g. xtc file) and the velocities from
> the gro file are used (hence also the total energy is affected). The
> potential energy terms should be virtually identical, although small
> differences may arise from the fact that coordinates in the xtc file are
> not exact.



David,

I'm rerunning with a trr file containing velocities. And the potential
energies are different, as are total energies, etc. Additionally, the
calculations I'm looking at at the moment are free energy calculations, so
the dv/dlambda values are different as well.

Thanks,
David


>
> > Could you perhaps try the same experiment on your own system? Just try
> > mdrun -rerun of a trajectory by taking your original mdp file, setting
> > nstlim=1, and then generating a tpr file with your original system's
> > starting coordinates and processing your original trajectory with that
> > tpr file and -rerun. Then take your original system's ending
> > coordinates (in my case, I've got a protein which I'm running for 1
> > ns) and repeat the same process. Compare the energies written out by
> > the two and see if they're different; I'm getting, as I said,
> > differences of several percent.
> >
> > Or could you maybe check the source code for the -rerun option or
> > something?
> >
> > I can even send you the system I'm working on if you want. I'm
> > actually running a very simple system with just toluene in water and
> > I'm seeing this difference; it's rather concerning, plus it's making
> > it so I can't do the data analysis I need.
> >
> > Thanks,
> > David
> >
> >
> >
> > On 10/24/05, David <spoel at xray.bmc.uu.se> wrote:
> > On Mon, 2005-10-24 at 14:43 -0700, David Mobley wrote:
> > > Dear all,
> > >
> > > I'm hoping someone can explain more thoroughly to me how
> > exactly mdrun
> > > -rerun works (Gromacs 3.2.1). Based on the manual, it looks
> > like it
> > > should be re-analyzing the trajectory I pass it. Thus, if I
> > rerun the
> > > same trajectory (with nstlist=1), I should get back the same
> > energies,
> > > etc.
> > >
> > > I've tried this several times on a particular system where I
> > need to
> > > reprocess some trajectories, and instead, the output depends
> > on what
> > > starting .gro file I use when I run grompp, not just on
> > the .top
> > > and .trr files that I use for mdrun -rerun. Exactly what
> > role does
> > > the .gro file play?
> > >
> > > For example, if I reprocess a production trajectory after
> > running
> > > grompp on a .gro file that I used for equilibration, my
> > reprocessed
> > > energies are several percent different than they were when I
> > first ran
> > > the trajectory. On the other hand, if I reprocess the
> > trajectory after
> > > running grompp with a .gro file corresponding to the final
> > frame of my
> > > original production trajectory, I get exactly the same
> > energies as I
> > > originally got.
> > >
> > > So what role is the .gro file playing in the analysis? Why
> > are my
> > > results when I reprocess the trajectory dependent on the
> > initial
> > > coordinates used for grompp?
> >
> > It should not matter at all. The only difference in energies
> > with the
> > original trajectory could be cause by chaning the tpr file. In
> > principle
> > it is possible to give a completely different tpr file (e.g.
> > for another
> > protein) and do mdrun -rerun (as long as the number of atoms
> > are the
> > same). Thus you have to take care that your new tpr file makes
> > sense. I
> > suggest you try
> > gmxcheck -s1 old.tpr -s2 new.tpr | more
> > to inspect the differences.
> >
> >
> >
> > >
> > > Thanks,
> > > David Mobley
> > > Dill group
> > > UCSF
> > > _______________________________________________
> > > gmx-users mailing list
> > > gmx-users at gromacs.org
> > > http://www.gromacs.org/mailman/listinfo/gmx-users
> > > Please don't post (un)subscribe requests to the list. Use
> > the
> > > www interface or send it to gmx-users-request at gromacs.org.
> > --
> > David.
> > ________________________________________________________________________
> > David van der Spoel, PhD, Assoc. Prof., Molecular Biophysics
> > group,
> > Dept. of Cell and Molecular Biology, Uppsala University.
> > Husargatan 3, Box 596, 75124 Uppsala, Sweden
> > phone: 46 18 471 4205 fax: 46 18 511 755
> > spoel at xray.bmc.uu.se spoel at gromacs.org
> > http://xray.bmc.uu.se/~spoel
> > ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> > ++++++++++
> >
> > _______________________________________________
> > gmx-users mailing list
> > gmx-users at gromacs.org
> > http://www.gromacs.org/mailman/listinfo/gmx-users
> > Please don't post (un)subscribe requests to the list. Use the
> > www interface or send it to gmx-users-request at gromacs.org.
> >
> > _______________________________________________
> > gmx-users mailing list
> > gmx-users at gromacs.org
> > http://www.gromacs.org/mailman/listinfo/gmx-users
> > Please don't post (un)subscribe requests to the list. Use the
> > www interface or send it to gmx-users-request at gromacs.org.
> --
> David.
> ________________________________________________________________________
> David van der Spoel, PhD, Assoc. Prof., Molecular Biophysics group,
> Dept. of Cell and Molecular Biology, Uppsala University.
> Husargatan 3, Box 596, 75124 Uppsala, Sweden
> phone: 46 18 471 4205 fax: 46 18 511 755
> spoel at xray.bmc.uu.se spoel at gromacs.org http://xray.bmc.uu.se/~spoel
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20051025/3067a191/attachment.html>