Re: [cnsbb] Can I refine 4-fold twinned data in cns?
- Hi Christine,
maybe this is worth looking at if you want to solve it through CNS;
/Acta Cryst./ (2005). D*61*, 613-621
/Escherichia coli/ MltA: MAD phasing and refinement of a
tetartohedrally twinned protein crystal structure
The authors adapted the CNS 'data_detwin' module to account for the
tetartohedral twinning. The modified files can be found in the IUCr
electronic archive (http://journals.iucr.org/services/earchive.html ;
This paper describes something similar and uses the files from the above
/Acta Cryst./ (2007). D*63*, 206-220
Structure of triosephosphate isomerase (TIM) from
This is a detwinning strategy however, and refining against the twinned
experimental data with the correct twin operator(s) is preferable. I
don't know how Refmac5 handles this, it could be the easier option though.
Unit for Structural Biology & Biophysics
Lab for Protein Biochemistry and Biomolecular Engineering
Department of Biochemistry & Microbiology
Ghent University, Belgium
Kay Diederichs schreef:
> Lukacs, Christine schrieb:
>> Hi –
>> I have a dataset that seems to be perfectly twinned – 4 times.
>> Long story short:
>> Apparent space group of P6122 gives MR solution, one molecule in AU,
>> beautiful ligand density; but R-factors stuck in the high 30s –low 40s.
>> I ran through the twinning manipulations (in ccp4) to try P6, P312,
>> and P321. Statistics show perfect twinning, but no luck with
>> So I went back to P3 - now I get a solution in P31, four molecules
>> in the AU, and four twin operators with fractions of 0.25 each. R
>> factors finally drop and the maps look good.
>> I would like to take this into CNX now. Can I specify more than one
>> twin operator in the various input files (make_cv_twin.inp,
>> detwin_perfect.inp, and anneal_twin.inp)?
>> If someone can help and would like more details, I am happy to take
>> this off-line.
> I'd say you can consider yourself lucky that the latest version of
> refmac5 can deal with tetartohedral twinning which you seem to have.
> AFAICT neither CNS nor phenix.refine support that kind of calculation.