On Thu, Dec 19, 2013 at 12:06 PM, Joseph Noel <noel@salk.edu> wrote:

I guess what I really would like to do is to derive experimental phases from the two wavelengths and the arrangement of Yb and S atoms in the structure. I would like to just provide the known coordinates for each and then derive phases that I can then use as restraints in the refinement of the structure at the first wavelength (longer wavelength). I know how to do this the old way but how do I do it in Phenix.

Extract the Yb and S atoms into a separate PDB file (use PDBTools for this, keeping only "element Yb or element S", and use this as input for AutoSol along with your anomalous data. �The output will include a file with Hendrickson-Lattman coefficients (calculated by SOLVE, if you have more than one wavelength). �You'll probably want to turn off model-building, since that's where most of the runtime goes.

Regarding difference map peaks around heavy atoms: anomalous refinement will definitely be helpful in this case, but I would probably save it for the very last round of refinement. �You won't be able to use the MLHL target but you can refine parameters (coordinates, B-factor, f' and f'') for just the Yb atoms and leave the protein alone. �(I doubt the rest of the model would change very much if you included it, though.) �But I'm curious what the map really looks like, because heavy atoms can cause weird effects, especially if your data are not truly "complete". �Supposedly the maximum-entropy treatment helps with this.

PS. Ensemble refinement can now use MLHL too (in nightly builds).

-Nat