Re: [ccp4bb]: strange NCS/refinement problem (twinning?)

[Bart Hazes <bhazes@ualberta.ca>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

On Fri, 13 Sep 2002, Dima Klenchin wrote:

> Hello all,
>
> I am facing a problem I am not sure how to address.
> how to approach (short of looking for a different xtal form).
>
> Good data set to 2A, indexes very well by P2 (67.947 76.813 98.441,
> beta=101.21) or P1 (alpha and gamma within 0.5 close to 90).
> Two molecules per cell. Scaled as P2, systematic absences indicate
> P21 very clearly.
>
> Molrep finds what appears to be a good solution readily. Problems
> start after that: When I restrain or constrain NCS during refinement,
> R free goes way up (R~30, Rfree > 40%). If I refine without NCS,
> R factors slip right away to 27/29 but this strange thing
> happens: one copy of the protein refines very well - low B factors,
> very good looking map, two ligands totalling >100 non-H atoms
> show up perfectly well on 1fofc map. A completely different
> story with another copy of the molecule: B factors are sky
> high and the map looks crappy with most of it being probably
> model bias. Matthews coefficient is 6.1 with just one molecule...
>
> At this point I have tried a lot of things hoping to find
> an error in previous steps and nothing shows up. Can this be
> twinned crystal? Yates server does not appear to think so.
> Short of trying to find a new well-diffracting crystal form,
> is there a reasonale solution?
>
> Thanks so much for any help,
>
> Dima

One possibility is that your two molecules are very similar but one has low
B-values and the other high B-values due to differences in packing
interactions. If your NCS restraints force the molecules to have similar
B-factors your Rfree will be going up. If you leave out the NCS restraints
B-values will shoot up for the ill-defined model and the density quality will
be poor. This would be an very interesting case for TLS refinement with
Refmac. Define each monomer as a TLS group to refine the B-values for them and
then refine the coordinates with NCS applied but keeping the B-values fixed.
The density for the poor model will probably never be pretty but it will give
you the lowest R/Rfree and you can then just look at the well-defined monomer
for your structural analysis.

Bart

===============================================================================

Assistant Professor
Dept. of Medical Microbiology & Immunology
University of Alberta
1-15 Medical Sciences Building
Edmonton, Alberta, T6G 2H7, Canada
phone:	1-780-492-0042
fax:	1-780-492-7521

===============================================================================