[gmx-developers] Re: trjconv -fit rot+trans -pbc broken, was: trjconv defaults

Thu Nov 5 00:20:03 CET 2009

Hi Tsjerk

On Wed, Nov 4, 2009 at 5:17 PM, Tsjerk Wassenaar <tsjerkw at gmail.com> wrote:

> Doing PBC stuff first and fitting later is still the only solution.
>

Well it is more of a work-around then a solution and it does not work at all
if the analysis tool, which is run after the fit, needs to do pbc.

> Somewhere in the archive there is a discussion digging further on this
> problem, but with exactly the same conclusion. There will still be a
> difficulty with visualization though, if the box is to be rotated. The
> PDB format does not allow inclusion of such arbitrary triclinic cells.
> They can't be captured in a CRYST1 record, which has vector lengths
> and angles, but not orientation.
>

Yes. It can't be fixed with PDB. But it can be fixed with the gromacs
formats because they have all 9 elements. Only at the moment gromacs only
supports  v1(y)=v1(z)=v2(z)=0 as you pointed out.

For making visualization work correctly the molfile plugin api (used by VMD
and Pymol) has to be extended too. Because at the moment molfile converts
the v1,v2,v3 vectors stored in GRO,TRR,XTC into the vector length and the 3
angles. Thus even if GROMACS would supported rotated box vectors vmd and
pymol would not read them correctly.

But at least the Gromacs analysis tools would do correct pbc after fitting.
And maybe we can convince the molfile people to extend the molfile api to
include full box vector description.

Roland

> Cheers,
>
> Tsjerk
>
> On Wed, Nov 4, 2009 at 8:35 PM, Roland Schulz <roland at utk.edu> wrote:
> > Hi,
> >
> > sorry small correction to my last email:
> >
> > It is OK to do pbc first and then do trjconv -fit trans+rot. Different to
> > what I wrote earlier it is correct that the rotation moves atoms out of
> the
> > box because it is the wrong box (their are still in the correct box). To
> > move them correctly in the translational shifted box you should do:
> > trjconv -fit translation -pbc {some-pbc}
> > trjconv -fit rot+trans
> >
> > This guarantees that the pbc treatment is correct in the original box.
> >
> > But the problem persists if you need pbc treatment in the
> > tools/visualization you use after trjconv. Since the written box vectors
> are
> > not rotated any pbc treatment applied to that trajectory will be wrong.
> >
> > So I still think the only real solution is the one proposed by Tsjerk. I
> > just wanted to add that there is a partial work-around.
> >
> > Roland
> >
> > On Wed, Nov 4, 2009 at 2:00 PM, Roland Schulz <roland at utk.edu> wrote:
> >>
> >> Hi,
> >>
> >> I'm replaying here to a year old thread because I found Tsjerk desribing
> a
> >> problem I'm currently having and I think no fix is so far planned.
> >>
> >> As he describes below if you do a fit then a later pbc correction is
> wrong
> >> because the box vectors don't get rotated together with the system.
> >>
> >> But this problem is not only introduced by changed defaults (as -ur
> >> compact) but is a problem with any -ur.
> >>
> >> Whenever one does a trjconv -fit rot+trans -pbc [any pbc] the pbc
> >> correction is wrong because the box vectors are wrong.
> >>
> >> Doing the pbc before is not a work-around because the rotation might
> move
> >> molecules out of the box again. Also first doing the fit and the calling
> >> trjconv again for -pbc does not work because the saved box vectors are
> the
> >> wrong (unrotated) ones and thus the pbc still are wrong.
> >>
> >> The only fix I see is Tsjerk's solution to also rotate the box vectors.
> Or
> >> is there any other solution?
> >>
> >> Roland
> >>
> >>
> >> On Fri, Oct 17, 2008 at 4:02 AM, Tsjerk Wassenaar <tsjerkw at gmail.com>
> >> wrote:
> >>>
> >>> Hi,
> >>>
> >>> No!
> >>>
> >>> I was already typing that I was in favour of setting -pbc to whole by
> >>> default, but not -ur compact, when the following struck me: Setting
> >>> these to defaults will screw up any trajectory/pdb which has before
> >>> been processed with options causing discrepancy between configuration
> >>> and pbc. In normal words, whenever you've performed a fit, you'll only
> >>> end up with garbled output running trjconv again.
> >>>
> >>> The solution to that, and I mentioned that to Erik at the workshop,
> >>> would be to have all fitting and other reorienting procedures also
> >>> perform on the box. That's a minor change, but then again, we have to
> >>> get rid of the 'box too skewed' error in analysis tools and be
> >>> prepared to use all nine numbers in the box matrix for the definition
> >>> of the PBC. Note that this doesn't work for PDB files, and there is no
> >>> work-around.
> >>>
> >>> In the margin of that, I also disfavour having the default for -ur set
> >>> to compact, because it will give unnecessary computational overhead
> >>> when processing trajectory files for whatever. Resetting to a
> >>> rectangular brick (or triclinic cell) on the other hand, is trivial.
> >>>
> >>> Cheers,
> >>>
> >>> Tsjerk
> >>>
> >>> On Fri, Oct 17, 2008 at 10:19 AM, David van der Spoel
> >>> <spoel at xray.bmc.uu.se> wrote:
> >>> > Hi,
> >>> >
> >>> > shall we set the trjconv defaults to -pbc mol -ur compact ?
> >>> > Even I get fooled when dumping a pdb file and suddenly noticing that
> it
> >>> > is
> >>> > broken.
> >>> >
> >>> > Cheers,
> >>> > --
> >>> > David.
> >>> >
> >>> >
> ________________________________________________________________________
> >>> > David van der Spoel, PhD, Professor of Biology
> >>> > 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://folding.bmc.uu.se
> >>> >
> >>> >
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> >>> > _______________________________________________
> >>> > gmx-developers mailing list
> >>> > gmx-developers at gromacs.org
> >>> > http://www.gromacs.org/mailman/listinfo/gmx-developers
> >>> > Please don't post (un)subscribe requests to the list. Use the www
> >>> > interface
> >>> > or send it to gmx-developers-request at gromacs.org.
> >>> >
> >>>
> >>>
> >>>
> >>> --
> >>> Tsjerk A. Wassenaar, Ph.D.
> >>> Junior UD (post-doc)
> >>> Biomolecular NMR, Bijvoet Center
> >>> Utrecht University
> >>> Padualaan 8
> >>> 3584 CH Utrecht
> >>> The Netherlands
> >>> P: +31-30-2539931
> >>> F: +31-30-2537623
> >>> _______________________________________________
> >>> gmx-developers mailing list
> >>> gmx-developers at gromacs.org
> >>> http://www.gromacs.org/mailman/listinfo/gmx-developers
> >>> Please don't post (un)subscribe requests to the list. Use the
> >>> www interface or send it to gmx-developers-request at gromacs.org.
> >>
> >>
> >>
> >> --
> >> ORNL/UT Center for Molecular Biophysics cmb.ornl.gov
> >> 865-241-1537, ORNL PO BOX 2008 MS6309
> >
> >
> >
> > --
> > ORNL/UT Center for Molecular Biophysics cmb.ornl.gov
> > 865-241-1537, ORNL PO BOX 2008 MS6309
> >
> > _______________________________________________
> > gmx-developers mailing list
> > gmx-developers at gromacs.org
> > http://lists.gromacs.org/mailman/listinfo/gmx-developers
> > Please don't post (un)subscribe requests to the list. Use the
> > www interface or send it to gmx-developers-request at gromacs.org.
> >
>
>
>
> --
> Tsjerk A. Wassenaar, Ph.D.
> Junior UD (post-doc)
> Biomolecular NMR, Bijvoet Center
> Utrecht University
> Padualaan 8
> 3584 CH Utrecht
> The Netherlands
> P: +31-30-2539931
> F: +31-30-2537623
> _______________________________________________
> gmx-developers mailing list
> gmx-developers at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-developers
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-developers-request at gromacs.org.
>

-- 
ORNL/UT Center for Molecular Biophysics cmb.ornl.gov
865-241-1537, ORNL PO BOX 2008 MS6309
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20091104/94788951/attachment.html>