[phenixbb] refinement with twinning

[Pavel Afonine]

  Hi Alex,

did you generate free-R flags using phenix.refine? If not, then try this:

- create free-R flags from scratch using phenix.refine;
- run ~10 refinement macro-cycles with the new flags. If the R-factors 
are similar, then this is what it is.

How the maps look like?

Have a look at page #123 here:
http://cci.lbl.gov/~afonine/tmp2/PavelAfonine_PHENIX.pdf

Let me know if you have questions.

Pavel.


On 2/16/11 11:52 AM, alexander.singer at utoronto.ca wrote:
> Hi -- I have a twinned data set -- approximate twinning fraction of 
> 0.4.  I decided to try to try twinned refinement using phenix.refine 
> in what would seem to be the appropriate manner -- from the "All 
> Parameters" menu, I select twinning, hit "Run Xtriage now" on my mtz 
> file, it gives me the input to the "Twin Law", I keep the rest of the 
> values the same (gave no Anisotropy correction), hit OK, then run 3 
> cycles of refinement.  The only thing is that the Rwork/Rfree values 
> are really very good, almost too good to be true -- for my 1.7 A 
> refinement, I have values of 12.4 and 14.8%, respectively for Rwork 
> and Rfree. Is there anything that I may be doing wrong yielding R 
> values which are artifically too low?
>
> Thanks for your help
>
> Alex
>