There are two different but equally good ways of indexing the reflections in space group I213, so there is a 50% chance that your new solution and the refined structure will be indexed differently, in which case the phases will not agree. Also you can easily improve the experimental phases from shelxe by using the beta-test version that does iterative main-chain tracing (available by sending me an email while I am still developing it).
George Prof. George M. Sheldrick FRS Dept. Structural Chemistry, University of Goettingen, Tammannstr. 4, D37077 Goettingen, Germany Tel. +49-551-39-3021 or -3068 Fax. +49-551-39-22582 On Mon, 9 Aug 2010, wtempel wrote: > Dear colleagues, > > here is one cunning plan: > > to quickly evaluate the anomalous signal of a test data set with a > non-interactive script that: > 1. solves the structure using SAD > 2. does some solvent flattening > 3. compares the resulting phases against calculated phases from a > refined, isomorphous structure. Generates some "global" measure of > phase error (deviation from refined, calculated phases). > > Not very original, but still failing in my hands. > > 1., 2. (shelxc/d/e) - check. > 3. a) take model amplitudes, phases and weights (F, PHIC, FOM(C)) from > a refmac MTZ file - check. > 3. b) cad PHI, FOM from the shelxe output with the selected columns > from the refmac MTZ - check. > 3. c) display reasonable maps using either of F-PHIC-FOM(C) or > F-PHI-FOM combinations - check. > 3. d) > > cphasematch -mtzin cad_ori.mtz -mtzout phasematch.mtz \ > -colin-fo "/*/*/[F,SIGF]" -colin-phifom-1 "/*/*/[PHIC,FOM]" \ > -colin-phifom-2 "/*/*/[PHI_ori,FOM_ori]" > > Now, I did expect to see a reasonable map using the F- > phasematch.Phi_fom.phi-phasematch.Phi_fom.fom combination that would > resemble F-PHI-FOM, but superimposed onto F-PHIC-FOM(C). But the F- > phasematch.Phi_fom.phi-phasematch.Phi_fom.fom map does not resemble > the protein structure at all. > > It is slowly coming to me that I must be doing this the wrong way. Can > anyone spot the problem? By the way, the crystal is cubic insulin > (a=78 A.) I would be happy to send out the data, the phasematch.mtz is > just 140 kB in size. > > Many thanks, > Wolfram Tempel > >
