[phenixbb] questions to my Phenix reifnemnt

Nathaniel Echols nechols at lbl.gov
Fri Dec 2 17:17:38 PST 2011

On Fri, Dec 2, 2011 at 3:36 PM, Dialing Pretty <hdc123hdc123 at yahoo.com> wrote:
> After coot rebuid,Igot a protein 3-D structure, and Phenix comprehensive
> validation demonstrated it is almost acceptable.Then I did the Phenix
> refinement , and I select all the strategies which suitable for my
> resolution, all the restraints and weights (default choice) suitable for my
> resolution,  all the other options suitable for me. I optimized it 10
> rounds.However after I got the PDB output file and the mtz output file after
> refinment, analysis by Phenix comprehensive validation demonstrated the
> quality of the quality of the protein 3-D structure has been lowered
> significantly, which makes it unacceptable.

What are the RMS(bonds) and RMS(angles), and which version are you
using?  You probably just need to optimize the X-ray/stereochemistry
weight - there is an option for this in the second tab in the GUI (or
on the command line, optimize_xyz_weight=True).

> Will you please tell me how to process my refinement? In Phenix refine,
> which kinds of combination of strategy combination are allowable, which are
> not allowable, which kinds of combination of restraints and weight
> combination are allowable, which are not allowable, which kinds of
> combinations of option are allowable, which are not allowable, and why?

For strategies, the main restriction is on overlapping atom selections
for group_adp and individual_adp, plus if you are refining individual
anisotropic ADPs for any atoms, these may not be part of any TLS
group.  Several of the other methods (such as rotamer correction) may
not be compatible with twinned refinement (it's a bug, sorry).  You
probably shouldn't use both types of simulated annealing at once.
There are many other empirical rules that are dependent on resolution
and structure quality, which we don't have a good summary for yet.
There is also an incomplete list of frequently asked refinement
questions here:


If you email the log file to help at phenix-online.org, we may have
suggestions more tailored to your particular situation.  (You should
send it to that address, not the phenixbb, since it may contain
revealing information about your structure.)


More information about the phenixbb mailing list