Hi Florian,
I'm currently refining a small protein against 0.96A data using phenix.refine (1.6-486) from the GUI. Refinement without hydrogens yielded Rwork/free of 12.4/14.2. I then added H-atoms with phenix.reduce and switched to individual refinement of hydrogens which dropped the R-factors to 11.3/13.4.
- the drop in r-factors is expected. For review see: Acta Cryst. (2010). D66, 1153-1163 Joint X-ray and neutron refinement with phenix.refine P. V. Afonine, M. Mustyakimov, R. W. Grosse-Kunstleve, N. W. Moriarty, P. Langan and P. D. Adams - it's always a good idea to use H atoms in refinement, especially at this high resolution, and a must at ultra-high resolution. For definitions see: Acta Cryst. (2009). D65, 1283-1291 On the use of logarithmic scales for analysis of diffraction data A. Urzhumtsev, P. V. Afonine and P. D. Adams
After this refinement the occupancies of some hydrogens dropped to 0 which makes me a little concerned.
I recall even for Aldose Reductase refined at 0.66A resolution we could see only ~70-80% of all possible H atoms. The are (many) reasons for this. At 0.96A I would expect to see much less. So this is why individually refined occupancy of an H atoms may drop to 0.0.
I would like to ask whether it is possible in phenix to group the occupancies to keep them at the same value as the rest of the residue but still refine xyz and isotropic adp of the individual hydrogens?
It used to be possible in the past but I removed that functionality (and now thinking of restoring it back but for a different purpose -:) ). For now I would just stick to riding model as it seems to be more appropriate in this case. All the best! Pavel.