By the way, I have a related question for Pavel. When you guys were culling the PDB for structures, was there any attention paid to correct the ones that have residual B-factors in the PDB file?
Yes, it's taken care of automatically by phenix.model_vs_data: it automatically recognizes if a PDB file contains residual or total B-factors. Twinning is takes care of too. There will be paper about it soon. The results of phenix.model_vs_data run for "all" PDB entries forms the database for POLYGON.
I am asking this, because a few structures I solved recently to ~2 Angstrom resolution were all closer to the upper range of observed structures according to POLYGON, so I thought this could possibly be one reason.
If you tell me which structures concern you then I will check. Pavel.