Hi Joe,
If one would like to try different weights for anisotropic ADP refinement (ie fixing wxu) should one be focusing on numbers like - wxu = 199.878301 - instead of the usual low numbers such as - wxu = 0.25 - found during isotropic ADP refinement?
these values may be very different depending on many pretty general things such as data resolution, model quality, etc... So it's hard to tell what is a good/bad or typical value... I keep eye on X-ray/Geometry weight and it appears that the typical value is around 1.-20. or so, but it can be drastically different if you switch from ML target to something else, like twin LS target.
While optimize X-ray/ADP weight is useful in phenix, I am still getting anistropic ADP values that are a bit outside what one might expect at 1.5 A compared to the PDB. For the latter, I check things using Parvati (http://skuld.bmsc.washington.edu/parvati/).
Are they nonsensical (physically impossible/wrong) or just different from "what is typically observed"? If the latter is the case then, provided that everything else is good and makes sense, I wouldn't worry about it too much since an outlier does not necessarily means wrong. If the first is the case then I would set a stricter weight for ADP refinement manually (rather than letting it get optimized using optimize_wxu=true). Let me know if you need any help with this. Good luck! Pavel.