On 09/03/2015 10:58 PM, Keller, Jacob wrote: > > Even SAD might be workable when the twin operator doesn't > superimpose all Bijvoet mates on each other (I think this is possible). I have encountered the situation only once, but in that instance the presence of substantial merohedral twinning did not interfere at all with MAD phasing. Solve/Resolve found 5 of 6 Se sites and auto-traced a large portion of the structure in the initial map. Refinement against the peak wavelength (so essentially the SAD case) eventually yielded R = 0.135 for 2A data.[*] The refined twin fraction was 0.39. The PDB code is 2B4W if you want to play with the data. An earlier example of direct experimental phasing in the presence of strong merohedral twinning was reported by Rudolph et al (2003) Acta Cryst. D59:290-8. doi:10.1107/S0907444902021704 PDB 1L2H Probably there are more recent examples as well but I haven't been tracking them. Ethan [*] Beware that the R values from twin refinement are always lower that you would otherwise expect. -- Ethan A Merritt Biomolecular Structure Center, K-428 Health Sciences Bldg MS 357742, University of Washington, Seattle 98195-7742