Hi Irina,

meaningfulness of what you are trying to do mostly depends on data resolution. If data resolution is better than 3A then it is all good to refine individual isotropic B factors of all atoms. Since Os is electron-rich compared to the rest it may be good to refine anisotropic B for Os only (isotropic for the rest). I don't think you need to refine group B. Yes, you need to refine all - coordinates, B and likely occupancies of Os. Error in one of these parameters will go into other parameters. R-factors 25/34 is not good so I guess you are using incorrect refinement strategy and/or model parametrization.

If you send me model and data and ligand cif files (if any) off-list I might be able to suggest something that may work better.

A good approach is to run with all defaults first, and then if not satisfied with results try to find a better way.

Good luck,
Pavel

On 10/11/15 09:18, Irina PROKHOROVA wrote:

Hello,

I have many Osmium ions in my structure. They have 6 coordinated ammonium groups and I would like to refine the B-factors for them: one B-factor for Os in each residue and one B-factor for all 6 ammonium groups in each residue.

For this I have chosen *group_adp with following syntax

    adp {
     group_adp_refinement_mode = one_adp_group_per_residue \
                                  two_adp_groups_per_residue *group_selection
      group = chain O and element Os
      group = chain O and element N
    }

I wanted to refine only B-factor for Osmiums in this refinement. To avoid the bias from previously refined values of B-factor I presetted B-factors for all Osmiums to 50.

As a result I got the change of B-factors for all residues in the model to the same value (which probably means that all model was refined as one group) and R-free increased from 0.25 to 0.34.

What should be the correct syntax for this refinement?

Should I also refine ions coordinates for a good refinement of B-factors?

Irina