[phenixbb] Defining heavy atom sites for AutoSol

[Randy J. Read]

Dear Kyle,

Just to amplify what Tom said, usually if Phaser adds sites, they represent 
something real, e.g. another (perhaps unexpected) anomalous scatterer (S, 
Cl, metal, depending on the wavelength) or a split site arising from static 
disorder. If the sites are real, deleting them can degrade your phase 
quality significantly.

However, I've seen a few data sets where there's some noise in the data 
that leads to false sites, so it wouldn't hurt to do a run on the side 
where you limit the number of sites.

All the best,

Randy

On Jul 22 2010, Kyle Dolan wrote:

>Hi all,
>
>I used AutoSol to generate SAD phases for a protein after completing a
>partial MR model. My strategy worked, but I noticed there were more heavy
>atom sites in the solution than I was expecting. Having continued to build
>and refine the structure, I am fairly certain that I know which sites are
>the real sites and which ones are noise. I would like to use the real sites
>to reproduce the experimental phases in AutoSol or another Phenix module to
>perhaps improve the map quality (essentially running just SOLVE or Phaser
>without HYSS or model-building). What is the right approach? I am guessing
>something like this:
>
>phenix.autosol data.sca sequence.dat sites_file=sites.pdb build=False
>
>Cheers,
>Kyle
>
>Kyle T. Dolan
>Department of Biochemistry and Molecular Biology
>The University of Chicago
>ktd at uchicago.edu
>