VMD-L Mailing List

From: Andrew Dalke (dalke_at_bioreason.com)
Date: Wed Aug 19 1998 - 13:49:12 CDT

Christoph Schneider <csc_at_imb-jena.de> said:
> I wonder that the fitting option in vmd is unnatural.
> When I compare 2 molecules (dinucleotides, same names/number of atoms),
> which should be almost the same in card. coord. (rmsd ~ 0.00x)
> it makes a great difference if I compare
> ***
> heavy atoms (rmsd ~ 3.4 A) NOT TRUE!! and bad superposition or
> backbone atoms (rmsd ~ 0.004 A) TRUE!! and exact superposition.
> all atoms (rmsd ~ 2.38 A) NOT TRUE!!
> ***
> But the result should almost be the same, I think.
> I think this is an important tool for checking equilibration runs and
> should not be underestimated!!

The underlying code is nearly identical, except for the selection
of which atoms to compare, and since the "backbone atoms" comparison
worked, I doubt there is a problem with the code.
  If this was a protein I would suggest that you have something like
a sidechain that was flipped between the two molecules. (For example,
if the tyrisine ring rotated 180 degrees it gives a large rmsd
even though they look almost the same.) But I don't think
nucleotides have the same problems since there's no symmetry
in the bases (unless I've forgotten something important :)
  There's also no good way to give hydrogen positions so they can
differ from molecule to molecule in a similar manner, but that
doesn't explain the problem with "heavy atoms".

  Here's what you can do to see if there are a few atoms that
are different between one molecule and the other. It is a procedure
that draws a green line between an atom of one molecule and the
corresponding atom on the other molecule.

proc connect_atoms {mol1 mol2} {
 set sel1 [atomselect $mol1 all]
 set sel2 [atomselect $mol2 all]
 set pos1 [$sel1 get {x y z}]
 set pos2 [$sel2 get {x y z}]
 draw color green
 foreach xyz1 $pos1 xyz2 $pos2 {
  draw line $xyz1 $xyz2
 }
}

Load the two molecules, get their molecule id (eg, 0 and 1) and
run "connect_atoms 0 1".

The line isn't all that thick so you might want to replace it with
a cylinder of radius 0.1 or so, or turn off the molecule displays.

This should help you find out if some atoms are displayed from others
by a not-small amount.

                                                Andrew Dalke
                                                dalke_at_bioreason.com