The phasing of electron diffraction data poses considerable problems for traditional direct or heavy atom crystallographic methods since the data are incomplete or at less than atomic resolution (i.e. 1.1Å), and subject to errors arising from dynamical scattering effects. The Bricogne formalism for phasing diffraction data using maximum entropy (ME) and Bayesian methods has proved especially useful in these situations1 since this formalism is stable irrespective of data resolution and completeness, and is robust with respect to errors on the measured diffraction intensities.
Successes with this formalism this include a wide range of structures:
(1) The ab initio phasing of diketopiperazine C4H6N2O2 (Fig 1).
(2) The structure solution of CuCl2.3Cu(OH)2 (Fig 2).
(3) The ab initio phasing of two 2-D membrane data sets at ca. 6Å resolution without the use of image phases (Halorhodopsin and Omp F Porin4 - Figs 3 and 4). In addition, we have had some success in phasing the 3-D data of bacteriorhodopsin at 6Å.