I have an unusual problem in which my enzyme (a dimer) is covalently bound to its own C-terminus and so is daisy-chained across the crystal lattice. In the past, I have encountered and fixed this problem by extracting the covalent part of the molecule and re-phasing using that as the molecular replacement model. In this case, I am not able to due so. When I input the restraints, it either crashes refine or attempts to impose them on the corresponding residues within the ASU (understandable) but they are quite distal and this leads to "interesting" refinement results. Is there a good way of making phenix.refine do what I want? Or perhaps a workaround that I haven't considered yet?