but how does phenix.refine handels metal-to-metal interactions? How should one set the refinement of a metallo protein with a two metals centre, in order not to bias the final result (metal to metal distances) ?
Roughly, there are two possible approaches: - manually create a .cif file, maybe based on the one you got from phenix.metal_coordination - follow the examples in the "Definition of custom bonds and angles" section of the phenix.refine documentation (https://www.phenix-online.org/documentation/refinement.htm) Without understanding your situation very well, I recommend you try the simple custom bonds and angles first, while keeping the .cif file you got from phenix.metal_coordination . If this doesn't work out, please email me the relevant fragments from your pdb file (metals plus neighboring residues) and I'll take a closer look. Ralf