Hi Mike,
I have a question regarding the Wilson B information reported in REMARK 3 of a pdb file. I am preparing a pdb file for deposition, and I am wondering which is the most appropriate value to report as the Wilson B in REMARK 3?
I thought phenix.refine reports it... I'm pretty sure I implemented it really while ago. If not, then it's a bug, and then just run phenix.model_vs_data model.pdb data.mtz and it will give it to you along with all other interesting statistics.
My inclination is to report the Wilson B for the full dataset calculated by XSCALE. This value differs slightly from the Wilson B value calculated by phenix.refine and printed to the refinement log file (XSCALE=27.6 vs. phenix.refine=19.8).
Phenix is supposed to do a better job in calculating Wilson B. Ask Peter Zwart for details or just have a look at relevant publication.
I assume this difference is due to phenix.refine using a low-resolution cutoff of ~20A.
phenix.refine does not use any resolution cutoffs (unless you asked it to do). All the best, Pavel