Re: [ccp4bb]: pseudo-merohedral twinning

[Ulrich Baumann <ulrich.baumann@ibc.unibe.ch>]

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


We had a similar case of pseudo-merohedral twinning in P2(1) (Fab
fragments) with a beta angle of about 90 deg emulating orthorhombic
symmetry. We picked the least twinned crystals (generally the smallest
one  :-( ), detwinned data with a local version of Andrew Leslie's DETWIN
(estimating twinning fraction from Yeates H and H^2 statistics) and solved
via molrep. Refinement was carried out using CNS which has an option to
treat twinned data (and does the detwinning job etc as well). Worked well.
Alternatively, use SHELX for refinement. It has the advantage of refining
the twinning fraction as well but in our case results were inferior to CNS
, e.g. the hapten electron density was much much worse. Also it is pretty
slow for larger structures but usually works quite well, too. Another
last resort is to detwin data and use Refmac but in my opinion this is
inferior and the maps generally show more model bias. Also the twinning
fraction must be significantly different from 0.5 to do the detwinning in
any case.

BTW: how does the cumulative intensity statistics output by TRUNCATE look?


Cheers,

Uli

***************************************************
**  Ulrich Baumann, Dpt. f. Chemie und Biochemie **
**  Universitaet Bern, Freiestrasse 3            **
**  CH-3012 Bern, Switzerland                    **
**  phone +41-(0)31-631-4320/4324, fax -4887     **
***************************************************

On Wed, 8 Aug 2001 TIMM_DAVID@Lilly.com wrote:

> Hello,  Are there any suggestions for refining structures using 
> pseudo-merohedral twinned data?  I have crystals in P21 with a=c, 
> beta=95.63 deg, 2-fold NCS, Rmerge = 0.07.  Refinement at 2.4 Ang hangs 
> with R-factors ~0.3 using data indexed h,k,l or reindexed l,-k,h.  Can 
> index with similar Rmerge in C-centered monoclinic and orthorhombic space 
> groups, but does not pack or refine.  Cheers,  Dave Timm
>