I'm trying to refine a structure containing ZN bound to four cysteine residues.

I have had trouble with Real-space refinement mangling the residues around this ZN.

At first, Real-space refinement tried to turn the cysteine into disulfides.

I tried switching this off with…

resname CYS and name SG

…in the disulphide bond exclusion selection string under the Model interpretation menu (there are no disulfides in the protein).

But the cysteine-zinc cluster continued to be mangled. Inspection of the .geo file shows that the SG-ZN distance is being restrained, but the CB–SG–ZN angle is not.

I checked the relevant dictionary file...

phenix-1.17.1-3660/modules/chem_data/mon_lib/list/mon_lib_list.cif

And the correct bond and angle restraints do seem to be present:

.

.

.

 ZN-CYS   1 ZN      2 SG        single       2.340    0.020

.

.

.

 ZN-CYS   1 ZN      2 SG      2 CB      109.000    3.000

.

.

.

Suggestions on how to fix this welcome.