Re: [ccp4bb]: A SAD case...

["Eleanor J. Dodson" <ccp4@ysbl.york.ac.uk>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

OnLineHelpForm wrote:
> 
> I am using ccp4 version  release-4_2.
> I am using ccp4i.
> I am using the redhat73 operating system.
> My compiler is: native
> I installed using compilesource
> The problem is as follows:
> 
> Dear all,
> 
> Perhaps some of you have come across a situation similar to this:  I have a fairly complete selenium MAD dataset (10 Se atoms in total), reduced and scaled with excellent statistics in P212121, with a decent anomalous signal.  Shake-and-bake finds all 10 peaks, with I/sigma(I) between 15.0-5.0, using the peak wavelength as a SAS dataset. When I refine those peaks with MLPHARE against the entire MAD dataset, the statistics look very good: phasing power of 1.1 and R-Cullis of 0.85 across all resolution bins (for the refinement I am using the strategy outlined by Ian Tickle's tutorial).  Only the real occupancies tend to a low value, 0.1, while the anomalous occupancies remain at 0.85.  Density modification (solvent content of 50%) seems to improve the matter bringing the figure of merit from 0.36 to 0.55 (both RESOLVE and DM). Even better, the map is fairly interpretable, and RESOLVE can trace almost 500f the protein as a polyalanine model.  But, when I inspect the fom-weigh!
te!
> d maps, all of the Se's are in the solvent!
> 
> So, the question is: How is it possible this lack of density around the peaks if the figure of merit is 0.55, the phasing power 1.0-1.1 and the R-Cullis 0.85?  Have I got to discard the Se positions calculated by Shake-and-Bake?  Where could the error be? (As for the scripts I use to produce maps, they are pretty simple and had always worked, so it cannot be just that.)
> 
> Any help would be highly appreciated!
> 
> I have done the following patches:
> No
> Please send replies to the bulletin board.

 But if you have traced a good bit of the model, you must know where the
Se atoms are?

 Maybe you are plotting Se used for phasing on one hand onto a map
phased on the other?

 the MAD phasing statistics are equally good for eith hand, but of
course one one gives an interpretable map. Maybe RESOLVE tries both and
you have not noticed it selecting the other?

 Eleanor