Lionel Mourey wrote:
> In the course of the crystal structure refinement of a protein
> covalently modified by a small molecule, it seems that the latter adopts
> alternate conformations which account for electron density and can
> easily be built by playing around dihedral angles. Then I thought that
> everything could be easily handled in X-PLOR, in the same way as for
> alternate side chain conformations (and indeed there are some in this
> structure that refined pretty well), i.e. using the "duplicate..." and
> "constraints inter ..." statements. Alas, refinement instead pushed the
> two conformers away and out of density as to minimize steric hindrance
> of too close atoms while preserving bonds lengths (thus, in their new
> position, the two conformers are still bound to the molecule).
I have had to refine a small molecule bound to a protein in two,
noninteracting,
overlapping modes. I positioned the molecules with O and then used only
positional refinement; simulated annealing didn't seem to deal very well
with
the overlapping regions of the two copies of the molecule.
If the "constraints inter" is phrased properly, refinement should behave as
if the
two conformers do not interact, in which case it would not minimize steric
hindrance. If you have something at the beginning of your file like
constraints
inter = (segid " ") (segid " ")
inter = (segid " " or segid "ALT1") (segid "ALT1")
inter = (segid " " or segid "ALT2") (segid "ALT2")
end
where you've set the segids of the two conformations to ALT1 and ALT2, then
the atoms
should not interact at all.
A tangentially related tip: What worked for me in terms of occpuancy
refinement was to
refine the occupancies of the two groups, then add them up (the sum != 1.0)
and scale the occupancies
so that they sum to 1.0.
xrefin {*occupancy refinement*}
optimize group
q=( segid "ALT1" and resname CPD ) {* Ligand A *}
q=( segid "ALT2" and resname CPD ) {* Ligand B *}
nstep=15
drop=1
end
end
{* This will make sure the sum of the two occupancies is 1.0 *}
vector show ave (q) (resname CPD AND resid 1)
evaluate($a=$RESULT)
vector show ave (q) (resname CPD AND resid 2)
evaluate($b=$RESULT)
evaluate ($t=$a+$b)
vector do (q=$a/$t) (resname CPD AND resid 1)
vector do (q=$b/$t) (resname CPD AND resid 2)
{* This will make sure the sum of the two occupancies is 1.0 *}
Good luck,
Y
--
_______________________________________________________________________
Yoram A. Puius Albert Einstein College of Medicine
6th year M.D.-Ph.D. Department of Biochemistry
mailto:puius at aecom.yu.edu 1300 Morris Park Avenue, Bronx, NY 10461
http://www.geocities.com/Athens/Forum/7504
_______________________________________________________________________
