Hi Jon,

try this:

- run phenix.refine with NCS restraints and use "optimize_wxc=true" (just out of curiosity, try w/o this option). Try these strategies to see which one works best: "strategy=individual_sites+group_adp+tls" or "strategy=individual_sites+individual_adp+tls" (don't forget to specify TLS groups. How? It is a separate question).

- run CNS or Refmac (whatever you like) using NCS constraints.

Compare Rwork, Rfree and Rfree-Rwork, map CC for each residue (phenix.real_space_correlation), and then you have a conclusive answer. Everything else are speculations -:)

Answering your question - yes, NCS constraints are in (my ?) to-do list.

I would appreciate if you share the results of the above experiment with me. And I'm open for further discussion.

All the best!
Pavel.

On 10/6/09 6:55 AM, Jonathan Grimes wrote:

Pavel Afonine wrote:
Hi,

phenix.refine model.pdb data.mtz main.ncs=true

will do refinement with NCS restraints. The NCS groups will be determined automatically. It may be important to check the automatic choice.
See phenix.refine manual for details:
http://phenix-online.org/documentation/refinement.htm#anch24

Also, at this resolution you may consider using TLS and/or group ADP refinement.

phenix.refine does not have NCS constrains, which I believe you do not need at this resolution.

Pavel.

Hi Pavel,

   Got to disagree with you on this one. Depends on your obs to params. If youve
   got 90% solvent then perhaps.......I would always favour reducing my params by
   a factor of 5 at 3.6Angs.

Jon

PS Any plans for coding up NCS constraints !!!   ;-)
-- 
Dr. Jonathan M. Grimes,   
University Research Lecturer
Division of Structural Biology
Wellcome Trust Centre for Human Genetics
University of Oxford
Roosevelt Drive,
Oxford OX3 7BN, UK

Email: [email protected], Web: www.strubi.ox.ac.uk 
Tel: (+44) - 1865 - 287561, FAX: (+44) - 1865 - 287547       
  
  
_______________________________________________
phenixbb mailing list
[email protected]
http://www.phenix-online.org/mailman/listinfo/phenixbb
  