[buster-discuss] Refine occupancy on two different polypeptide chains

Clemens Vonrhein vonrhein at globalphasing.com
Mon Jan 29 09:48:57 CET 2018

Dear Anna

On Fri, Jan 26, 2018 at 10:17:01AM +0100, Anna Suarez Larsson wrote:
> I am dealing with a pentameric structure composed of two different
> short (80 aa) polypeptide chains, A and B, that are 40%
> identical. In the crystal packing it looks to be no preferred
> orientation for the ring with respect to the two different
> proteins. I would like to combine the two chains in the same
> structure and refine with Buster to get the relative occupancy.

Yes, that should be doable.

> We know from gel electrophoresis of the two proteins coexpressed
> that the relation is 3:2 (A to B) between them, both in crystals and
> in solution. The crystals diffract to about 2.0 � resolution but the
> spots are streaky. The R-values when refining with only one of the
> polypeptide chains are much lower for A compared to B.


> My attempt to do the refinement in Buster using first pdb2occ and then run with the command -Gelly failed with the comment:
> Interpreting CONSTANT and FREE cards: 
>  *** ERROR found in interpreting atom specifier from CONSTANT card
>  *** ERROR specifier=FixOcc
>  *** ERROR does not match any atoms
>  *** ERROR *** found processing constant cards

Yes: we never anticipated that the complete model consists of atoms
that will get their occupancy refined - so the FixOcc set/group is
actually empty.

> How should I do to be able to proceed?

What happens if you remove that line (NOTE BUSTER_CONSTANT OCC FixOcc)
from your Gelly file?

> Is it possible to refine double occupancy on all the five
> polypeptide chains or is it required that I keep some of them with
> fixed occupancy?

You should be able to do that, although using pdb2occ might not be the
most elegant way: this tool is mainly intended to handle occupancy
refinement situations with some small(ish) localised items (alternate
conformation side-chains, partially occupied ligands, alternate loop
conformations etc).

Please let us know if you hit another issue.

Kind regards

Clemens & Andrew (for buster-develop)

More information about the buster-discuss mailing list