sulfate occupancy

Yoram Puius puius at
Thu Jan 8 10:40:33 EST 1998

Diao Jia-Sheng wrote:

> Hi, All:
>    I am refining a structure including a sulfate group.
>    I'd like to know the accurate occupancy of this sulfate group.
>    What I have done is as follows:


> 5. I have ever noticed that if I iterate the step 2 and step 3 alternatively,
>    it seems the values of the occupancies and b factors will always go down
>    together without convergement.

As you mentioned, they're really correlated.

> My questions: 1. How can I get the accurate evaluation for the ocupancies of
> the
> sulfate group?

Do you have a reason to believe the sulfate group does not have an occupancy of
~1.0?If you assume q=1.0, does the system behave happily?  With the typical
resolution of
protein crystallographic data, you may as well shave with Occam's Razor;  if
the simplest hypothesis is q=1.0 with sensible B-factors, I'd stick with that,

BTW, is the sulfate on a special position or anything?  That probably wouldn't
refine too well in any case.

> 2.  Should I do iterating process?  Maybe I should just regard the values
>  output from the first round of step 3 refinement as what I need ?

 I have done something similar once.  I had two ligands bound to a protein in
exclusive modes.  I basically did grouped occupancy refinement of both ligands,
re-set the occupancies to sum to 1.0, and then did *individual atom* B refinement.

Because of the constraint for the occupancy sums to be 1.0, plus the fact the data

was 1.9A resolution, it converged well.  If I didn't reset the occupancies to sum
to 1 each time, though, they would both move to funny values after a while --
they actually tended to run a little high.

So I think the iterative approach is fine, but if there's no reason to use it,
keep it
simpler and let q=1.0.

    - Y

Yoram A. Puius                Albert Einstein College of Medicine
6th year M.D.-Ph.D.           Department of Biochemistry
mailto:puius at     1300 Morris Park Avenue, Bronx, NY  10461

More information about the X-plor mailing list