In SHELXL this is handled by refining the ligand with "PART -1" and occupancies
fixed at 0.25, and refining the rest of the structure normally. The program ignores
interactions between symmetry generated atoms with negative PART numbers
when generating restraints, placing H-atoms etc. Positive PART numbers work
like the alternative locations A, B, C etc. in a PDB file but unfortunately the PDB
file has no equivalent to negative PART numbers. Solvent and other molecules
disordered on special positions are common in small molecule structures;
toluene just loves to sit on an inversion center.

George

On 02/08/2016 01:04 AM, Dale Tronrud wrote:

    My bad!  As Boaz Shaanan pointed out to me, I missed the "NCS" in
your subject line.  This means that my answer was not relevant to your
question.

    The source of the confusion surrounding your situation is that the
PDB file format does not easily allow one to describe exactly what your
model is.  What you really have is a tetramer with a single ligand at
its center, which breaks its symmetry and makes it a P1 particle.  This
entire particle sits in the crystal in a disordered fashion, since the
outside has four-fold symmetry.  Refinement programs are not going to
like four entire tetramers at 1/4 occupancy where nearly all the atoms
superimpose.

    Compounding the problem is the fact that the PDB file format does not
really specify what the relationship is between the "A" alternative
location of residue 1 and the "A" alternative location of residue 2.  I
would prefer to think that 1.A can bond to 2.A and have bad contacts
with 2.A but not to 2.B, but the format doesn't say that.  And if it
did, we would run out of letters for alt locs pretty fast.

    If you define your ligand to have four different names, the
refinement program will assume you mean that there are four residues in
the same universe and assume there are bad contacts between them.  If
you give the ligands four different names but specify different alt locs
(e.g. 1.A, 2.B, 3.C, and 4.D) the program still can't safely assume
there should be no bad contacts (because of the vague format
specification).  You should be able to give all four molecules the same
name but different alt locs (e.g. 1.A, 1.B, 1.C, and 1.D) and the bad
contacts should be ignored.  If there are alt locs for the surrounding
protein atoms that match the alt locs of your ligand you are back to the
format specification problem - it is not safe to assume that 123.A is in
the same universe as 1.A.

    When you find that the program is pushing apart atoms that you don't
think should be seeing each other, you can either change the names (and
there may be no workable naming solution) or use the tool in the
refinement program to tell it to ignore particular bad contacts.  Every
refinement program has such a tool.  The fact that you have to add this
ad hoc override is not a flaw in the program, it is a flaw in the file
format.

Dale Tronrud


On 2/7/2016 7:03 AM, Boaz Shaanan wrote:

Hi Dale,

Just wondering:  is the scenario your're describing also valid for 4-fold NCS axis (as is the case here), where the occupancies may not be as strictly related ? It sounds to me more  relevant to crystallographic 4-fold axis but I could well be wrong of course.

Cheers,

                 Boaz


Boaz Shaanan, Ph.D.
Dept. of Life Sciences
Ben-Gurion University of the Negev
Beer-Sheva 84105
Israel

E-mail: [log in to unmask]
Phone: 972-8-647-2220  Skype: boaz.shaanan
Fax:   972-8-647-2992 or 972-8-646-1710





________________________________________
From: CCP4 bulletin board [[log in to unmask]] on behalf of Dale Tronrud [[log in to unmask]]
Sent: Saturday, February 06, 2016 10:46 PM
To: [log in to unmask]
Subject: Re: [ccp4bb] Ligand at four fold NCS axis

    If you have a ligand sitting on a four-fold axis, let's call the four
residues 1, 2, 3, 4.  Residues 2, 3, and 4 are generated by the
refinement program using symmetry so the occupancy of 1 should be set to
0.25 to generate a total occupancy of 1.0.

    Now you say that residue 1 has four conformations.  Let's call them
1.A, 1.B, 1.C and 1.D.  The sum of the occupancies of these
conformations have to add up to the occupancy of 1.  If their
occupancies are equal you have to set them to 0.0625 (or 1/16).

    The refinement program will generate 2.A, 3.C and all the rest.

Dale Tronrud

On 2/6/2016 11:41 AM, Appu kumar wrote:

Dear CCp4 Users,
I am refining a tetrameric protein which has ligand placed at the center
axis. There is one tetramer in ASU.  I added the ligand with 0.25
occupancy and performed the refinement using both phenix and refmac.
Refinement leaves lots of green density around the ligand inferring that
there may be four different conformation of the ligand. When i made four
different conformation of ligand (each with 0.25 occupancy) and started
refinement, there is lot of clashes in ligand itself. Is there a ways to
refine the four different conformation of ligand with different
conformation with 0.25 occupancy avoiding clashes in ligands?

I appreciate your input, suggestions  on the refinement.

Thank you
Appu

--
Prof. George M. Sheldrick FRS
Dept. Structural Chemistry,
University of Goettingen,
Tammannstr. 4,
D37077 Goettingen, Germany
Tel. +49-551-39-33021 or -33068
Fax. +49-551-39-22582