[Chimera-users] PDB Coordinates

[Thomas Goddard]

Hi Albion,

  Elaine's suggestion to use the "Save relative to model" option when
saving coordinates of PDB models fit into maps is the right solution.

  Here are a few more details about coordinates in Chimera.  If you
open a PDB model in Chimera its atoms are positioned at Chimera
(x,y,z) coordinates just as they are specified in the file.  When you
rotate the model the atoms are now at different positions in Chimera
coordinates.  If you use File / Save PDB... without the "save relative
to" option you will see that the coordinates in the new file are
different from the original PDB file -- the file contains positions
for the rotated molecule.  If you open a PDB file and a density map,
rotate both together, save the PDB file, then restart Chimera and open
the original density map and the saved PDB file you'll see they no
longer have the same relative position.  The problem is that you saved
the rotated PDB model but you are using the original (unrotated)
density map.  There is no option to save the rotated density map, and
you don't really need that.  Saving a rotated map would require
resampling it at new grid positions because none of the map formats
include an option for specifying a rotation in the header.  You don't
want to resample because it reduces the quality of your map.  Instead
you use the "save relative to" option when saving the PDB file to say
to use coordinates relative to the orignal density map position, when
it was first opened in Chimera.  Effectively Chimera applies a rotation
and translation to the PDB coordinates that places your density map
back in its original position before saving the PDB atom coordinates.

  There is a bit more complexity to this.  Chimera remembers the
original PDB atom coordinates and the current rotation and translation
which positions the PDB model from all your cumulative mouse or
command rotations and translations.  If you start Chimera and open a
density map it appears in a standard orientation, x-axis horizontal,
y-axis vertical, and z-axis pointing out of the screen.  If instead
you start Chimera, open a PDB model, rotate it, now open the same
density map, it does not appear in the standard orientation!  Instead
it applies the rotation of the PDB model to the density map.  This is
a good behaviour.  It means no matter how you rotate and move your PDB
model before opening the density map, you end up with the PDB model
and density map having the same relative placement as if you hadn't
moved the PDB model at all before opening the density map.  But it can
get confusing.  If you first open two PDB models, the move one of them
with the other fixed (ie. deactivated), then open the density map,
which PDB model does the density map align with?  The two models now
have different rotation/translations relative to their original file
coordinates.  The answer is that the density map gets the
rotation/translation of the PDB file with the lower id number (shown
in Model Panel).  That might not be what you want.  Model Panel has a
"Transform As..." button that allows you to position the density map
using the same rotation/translation as the other PDB model if that is
what you want.

	Tom