Hello there, It's been a while we haven't chatted with anybody from the Phenix crew, I hope you are all doing fine! I'm writing because of an issue with this CryoEM structure at 2.5 angstroms I'm about to deposit. It is a myosin motor domain + part of the lever arm including one light chain. I've used RealSpaceRefine from Phenix 2.0 with very good results: very nice geometry and agreement with the map. The thing is, the lever arm and the light chain move a lot, so the contour in that part is much lower, ~50% the one of the motor domain, of course the map is also more poorly defined. And RealSpaceRefine attributes very low ADPs for the light chain (chain B), but the ones for the lever arm helix (chain A, same as the motor domain) are high, coherent with the difference in map quality. I'm considering "cheating" by putting them two as chain A to try to get this fixed (I'll need to undo this before deposition), but is there a more elegant way to sort this out? The PDB does not complain about the current model, but I'm thinking of readers looking at these ADPs and finding the ensemble really weird (basically a hot arm surrounded by a cold calmodulin-like domain, which is physically impossible). (To me it looks like a different normalization is being applied to the two chains because of the map quality difference, if this makes any sense.) I'm looking forward to your opinions, Carlos