[phenixbb] Question about unrestrained refinement

Pavel Afonine pafonine at lbl.gov
Tue Jan 10 12:15:50 PST 2012

Hi Louis,

> I had a question about unrestrained refinement. I have several data 
> sets (right near 1 A resolution) that are nearly fully refined, but to 
> finish I would like to run one round of unrestrained refinement. 

at might be a valid thing to do at subatomic resolution.

> At this point, I want to focus on unrestrained refinement of just the 
> protein, and not any waters. How would the syntax for refining just 
> the chains of the protein (and not water) look in the command line? 

> I also want to include anisotropic refinement of B-factors in this 
> last refinement run.

At this resolution you should always refine anisotropic ADP for all 
(including solvent) but H atoms.

> Should it look something like: 
> strategy=individual_sites+occupanices+individual_adp \ 
> adp.individual.anisotropic="not water and not element H" wc=0 wu=0

It depends how you run phenix.refine: using the GUI or typing commands 
from the command line or using parameter files. From the command line 
something like

phenix.refine model.pdb data.mtz sites.individual="(chain A and resseq 
1:10) or (chain C and resname ALA)"

will refine coordinates of atoms in residues from 1 to 10 in chain A, 
and coordinates of atoms in ALA residues of chain C. The coordinates of 
the other atoms will not be refined at all.

You can do the same for ADPs.

wc=0 will turn off geometry restraints (globally)
wu=0 will turn off ADP restraints (globally).

Probably, this is easier to do using a parameter file. For an example, 
see page 58 here:

Please let me know if you have any questions or problems with this.


More information about the phenixbb mailing list