[phenixbb] Improvement of low resolution MR phases

[Pavel Afonine]

Hi Tommi,

5-6 A is a tough resolution so don't expect things easy with any method 
or software package. Plus, methods (eg., density modification, 
refinement, model building) are typically optimized towards higher 
resolution. But with some luck and effort, you can make it work. Some 
suggestions that are appropriate to try at this resolution:

- if NCS is present, try NCS constraints vs restraints, and see what 
works best;

- optimize coordinate and B factor weight if Rfree-Rwork gap is overly 
large or geometry is poor;

- use all possible information (if available): reference model 
restraints, secondary structure restraints (with careful 
manually-verified annotation of SS), Ramachandran plot restraints;

- if model needs some substantial movement to fit the map, try Simulated 
Annealing;

- try density modification to see if you get a more interpretable map 
that can help you to build a better model (Tom can correct me, but I 
thought NCS averaging happens as part of DM).

This is all I can think of...

Pavel

> Hello all,
>
> I was wondering what would be current best protocols for trying to 
> improve maps/phases for low resolution
> MR / refinement solutions (5-6 Å resolution).
>
> High solvent content of course, dimeric 2:2 protein complex so some 
> NCS averaging and density modification
> I assume might help (related structures known, currently maps very 
> blobby, clear solution but I think should be able to do better...).
>
> Also, refinement suggestions welcome.
>
> Best route in phenix? So far worked initially with ccp4. Other 
> suggestions also welcome.
>
> Thanks!
> Tommi
>