[phenixbb] maps : sigma-a

Randy Read rjr27 at cam.ac.uk
Wed Oct 6 07:36:05 PDT 2010


Hi,

The application we're talking about here is using a model that hasn't seen all the data in refinement to decide how far there's real signal in the data that were measured.  We used that for the structure of 1bos (carbohydrate complex with Shiga-like toxin B-subunit), where the data should have been cut off at about 3.5A according to normal I/sigma criteria.  However, by including weak data to 2.8A (helped in no small measure by 20-fold NCS), we were able to get very clear density for the bound carbohydrate.  We validated the use of the weak data by taking the molecular replacement model before refinement and computing a SigmaA curve against all the data.  The SigmaA values were significantly above zero up to 2.8A (indicating a correlation with the observed data, and since we hadn't done any refinement against those data this confirms that there's really some useful signal there), but dropping off dramatically at higher resolution.  Axel has done similar things with even lower resolution structures where they really don't want to give up any information.

Something like this could be done with cross-validation data after refinement, but the SigmaA determination is much more reliable with more reflections, so if it can be done using all the data before refinement, the results will be clearer.

Regards,

Randy Read

On 6 Oct 2010, at 15:10, Pavel Afonine wrote:

> Hi,
> 
> isn't it better to use all data in refinement and expect that the ML target will down-weight those reflections that needs to? And if you have one single reflection at 1A resolution, while the data extends complete up to 2A resolution, you would use it up to 2A anyway? 
> Anyway, I hope Randy might comment on this...
> 
> Regardless, as I wrote yesterday I will add this functionality as soon as I get a chance.
> 
> Pavel.
>   
> 
> On 10/6/10 6:48 AM, crystallogrphy wrote:
>> 
>> Hi, Pavel:
>> I actually also need this function: sigma vs resolution, because I want to determine what is the real resolution of my crystal data. Randy Read and Axel Brunger had a paper talking about it. 
>> 
>> On Wed, Oct 6, 2010 at 12:47 AM, Pavel Afonine <pafonine at lbl.gov> wrote:
>>  Hi Bryan,
>> 
>> 
>> how might one start to go about getting phenix to automatically write out the value of sigma_A vs. resolution?
>> 
>> e.g. when making maps, to get that nice table in Randy Read's SIGMAA program output of ln(sigma_A) vs. resolution.
>> 
>> easy -:) Simply nobody asked for this so far so it's not there. I will add it within next few weeks and let you know once it is available. It's not computed at the moment (we use different parametrization) but I can reformulate it to get what you want.
>> 
>> May I ask you why you need it?
>> 
>> Thanks!
>> Pavel.
>> 
>> 
>> _______________________________________________
>> phenixbb mailing list
>> phenixbb at phenix-online.org
>> http://phenix-online.org/mailman/listinfo/phenixbb
>> 
>> 
>> _______________________________________________
>> phenixbb mailing list
>> phenixbb at phenix-online.org
>> http://phenix-online.org/mailman/listinfo/phenixbb
> 
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb

------
Randy J. Read
Department of Haematology, University of Cambridge
Cambridge Institute for Medical Research      Tel: + 44 1223 336500
Wellcome Trust/MRC Building                   Fax: + 44 1223 336827
Hills Road                                    E-mail: rjr27 at cam.ac.uk
Cambridge CB2 0XY, U.K.                       www-structmed.cimr.cam.ac.uk

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20101006/ccfd9e50/attachment-0002.htm>


More information about the phenixbb mailing list