[phenixbb] phasine error

[Pavel Afonine]

Hi Nat,

it is all published. Below are the papers that I used in phenix.refine 
to code maximum-likelihood target, phase and coordinate errors 
estimates. The complete list of relevant papers will include those by 
Randy Read, G. Bricogne, and others.

Alpha/Beta and phase error estimation algorithms, ML target formula:
V.Yu., Lunin & T.P., Skovoroda. Acta Cryst. (1995). A51, 880-887. 
"R-free likelihood-based estimates of errors for phases calculated from 
atomic models"

Formula for coordinate error estimation:
V.Y., Lunin, P.V. Afonine & A.G., Urzhumtsev. Acta Cryst. (2002). A58, 
270-282. "Likelihood-based refinement. I. Irremovable model errors"

P. Afonine, V.Y. Lunin & A. Urzhumtsev. J. Appl. Cryst. (2003). 36, 
158-159. "MLMF: least-squares approximation of likelihood-based 
refinement criteria"

Paper suggesting to use TEST reflections in Alpha/Beta estimation:
A.G. Urzhumtsev, T.P. Skovoroda & V.Y. Lunin. J. Appl. Cryst. (1996). 
29, 741-744. "A procedure compatible with X-PLOR for the calculation of 
electron-density maps weighted using an R-free-likelihood approach"

> they will influence the  
> final phases calculated in phenix.refine, but those will still be  
> biased towards the refined model.
>   

No, because the estimates are done using TEST reflections that are not 
supoosed to be used in any step of structure solution and refinement.

Pavel.


On 6/11/09 10:20 AM, Nathaniel Echols wrote:
> On Jun 10, 2009, at 9:18 PM, Pavel Afonine wrote:
>   
>> you can get it using phenix.refine. Just type from the command line:
>>
>> phenix.refine model.pdb data.mtz strategy=none  
>> main.number_of_macro_cycles=1
>>
>> and hit Enter. Look for the numbers towards the end of log file or  
>> output to the screen (after bulk solvent correction and scaling is  
>> reported done).
>>     
>
> I've never been able to understand the math involved, but isn't the  
> phase error calculated by phenix.refine is based on the estimates of  
> model error (and incompleteness) used for maximum likelihood  
> calculations, not an evaluation of the original experimental phases?   
> If experimental phase restraints were used (which is the default for  
> phenix.refine, and I assume AutoBuild too), they will influence the  
> final phases calculated in phenix.refine, but those will still be  
> biased towards the refined model.
>
> -------------------
> Nathaniel Echols
> Lawrence Berkeley Lab
> 510-486-5136
> NEchols at lbl.gov
>
>
>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://www.phenix-online.org/mailman/listinfo/phenixbb
>   
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20090611/0ecad8a9/attachment-0003.htm>