VMD-L Mailing List

From: Leonardo Trabuco (ltrabuco_at_ks.uiuc.edu)
Date: Sat Aug 25 2007 - 01:31:45 CDT

Dear Jim,

On Fri, Aug 24, 2007 at 04:58:51PM -0600, Jim Pfaendtner wrote:
> I am selecting 3 alpha carbons from one molecule and 3 alpha carbons
> from another molecule and I would like to align the two molecules
> based on the selection of these 6 atoms.

It is not uncommon that the alignment algorithm doesn't work well with
such a small number of atoms. You may want to try using 4 atoms instead
(you could create a dummy atom and set it to the geometric center of the
other 3 atoms); I believe that works better that using 3 atoms.

> [...]

> When I look at the resids in each of my selections, the selection is
> ordered from highest resid to lowest. I guess this has something to
> do with the way the atoms are searched in the pdb during atomselect,
> i.e., the order that I specify (selection 1), (selection 2),
> (selection 3) does not matter...

That's correct, it does not matter: the selection is ordered by
increasing index values, if you wish. This is an efficient and general
behavior. If the order mattered, what would be the order of the
following atom selection? :-)

atomselect top {index 10 or index 5 or index 10}

> I think frequently the alignment procedure is used on two identical
> proteins, so this is probably not an issue because the residue
> numbers are all the same.
>
> One idea I had was to perform all six atom selections separately and
> then form a concatenated atom selection in the order I want. Is this
> possible?

Nope.

> Does anyone else have a suggestion how to do this alignment?

Check out this message from a couple of days ago:
http://www.ks.uiuc.edu/Research/vmd/mailing_list/vmd-l/10172.html

I short, you can give an order parameter to 'measure fit'.

Cheers,
Leo 

-- 
Leonardo Trabuco, Ph.D. student
Theoretical and Computational Biophysics Group
University of Illinois at Urbana-Champaign