What is happening here is that Chimera's topology file parser is trying to
guess what the element of each atom is. Atom name is unreliable (every
carbon-alpha is named "CA", which makes it hard to distinguish between
calcium and carbon, for instance), as is atom type. Chimera tries to use
that information, but not very hard (i.e., if the atom type is not the same
name as the element, it tries the masses) -- I checked the source code to
figure out what was happening.
Typically, the most reliable way to determine the atomic number, by far, is
using the atomic mass (unless of course the atomic number is available
directly). So when you repartition the masses, Chimera starts assigning
the wrong elements to most of the atoms. This is why using the original
prmtop file works, but the repartitioned masses one does not.
The prmtop only recently started storing the atomic numbers directly, and
the Chimera parser has not been updated for that change. For Chimera
developers, that is the easiest fix.
All the best,
Jason
On Thu, Aug 20, 2015 at 9:03 PM, George Green <soyo.green.gmail.com> wrote:
> Thanks Adrian & Hai,
>
> So I guess if i just use the original (nornmal 2fs) topology file for all
> cpptraj, vmd and chimera analysis, there should be no problems.
>
> Glad the solution was that simple. Was anticipating a bit of a nightmare.
>
> Many thanks!
>
> On Fri, Aug 21, 2015 at 1:32 AM, Nhai <nhai.qn.gmail.com> wrote:
>
> > I think HMR topology affects the analysis requiring mass, like center of
> > mass. So it's better to use original top (which is easy to generate).
> >
> > Hai
> >
> > > On Aug 20, 2015, at 8:18 PM, George Green <soyo.green.gmail.com>
> wrote:
> > >
> > > yup, thought it would go that way. though, i was wondering whether this
> > is
> > > going to affect the analytical tools offered in 3rd party programs such
> > as
> > > chimera, pymol and vmd. This might make analysis complicated! Surely
> > > someone else must have had to deal with this issue?
> > >
> > >> On Fri, Aug 21, 2015 at 1:15 AM, Adrian Roitberg <roitberg.ufl.edu>
> > wrote:
> > >>
> > >> So, if you load the SAME coordinate file with the regular topology it
> > >> looks ok ?
> > >>
> > >> My only possible guess would be that chimera decided about distances,
> > >> etc based on the atomic mass and NOT on the atom type. If that is the
> > >> case, there is nothing you can do, except tell the chimera developers
> to
> > >> change that.
> > >>
> > >> adrian
> > >>
> > >>
> > >>> On 8/20/15 7:49 PM, George Green wrote:
> > >>> Hi,
> > >>>
> > >>> I set up up simulations to use hydrogen mass repartitioning and ran
> > some
> > >>> simulations with no apparent problems. However, on visualising with
> > >>> chimera, the protein does not show ribbons which makes visualisation
> > >>> difficult. Please see attached figure.
> > >>>
> > >>> The problem only occurs with the topology file created after feeding
> > into
> > >>> parmed to change masses. I guess the problem is due to atoms being
> > >> further
> > >>> apart than the defined distances in chimera? It would be good if
> > somebody
> > >>> could confirm that this is a known issue and not the result of
> > incorrect
> > >>> parametrisation? Also, if it is expected behaviour, is there any way
> to
> > >> get
> > >>> ribbon style visualisation back?
> > >>>
> > >>> many thanks
> > >>>
> > >>>
> > >>> _______________________________________________
> > >>> AMBER mailing list
> > >>> AMBER.ambermd.org
> > >>> http://lists.ambermd.org/mailman/listinfo/amber
> > >>
> > >> --
> > >> Dr. Adrian E. Roitberg
> > >> Professor.
> > >> Department of Chemistry
> > >> University of Florida
> > >> roitberg.ufl.edu
> > >> 352-392-6972
> > >>
> > >> _______________________________________________
> > >> 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
> >
> > _______________________________________________
> > 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
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Aug 21 2015 - 07:00:03 PDT
