DIRECT METHODS IN REAL AND RECIPROCAL SPACE. George M. Sheldrick, Universität Göttingen, Germany

Conventional direct methods that use probability relations to determine the phases of a limited number of reflections are computationally extremely efficient for small structures, but the chances of success decrease sharply as the number of independent atoms increases above about 200. They also require fairly complete data to atomic resolution. It appears that the size barrier has at last been broken by methods that iterate between real and reciprocal space, pioneered by the 'Shake and Bake' program developed by the Buffalo group (Miller et al., 1993). Given a powerful enough computer, much larger structures can be solved than were possible with pure reciprocal space direct methods. The success rate can be improved if slightly better than random starting phases are available, e.g. from automated Patterson interpretation (Sheldrick & Gould, 1995). Such a Patterson-based structure expansion enabled the ab initio solution of a small metalloprotein with about 840 non-hydrogen atoms in the asymmetric unit (Frazao et al., 1995). However these methods still require data to atomic resolution, which in practice means about 1.2 Å.

The next breakthrough will probably be the more active use of general structural knowledge in the real-space part of these procedures, rather than simply peak picking; it is possible that this will lead to a relaxation of the atomic resolution requirement. This talk will review recent progress towards the ab initio solution of both small-moiety and macromolecular structures.

C. Frazao, C.M. Soares, M.A. Carrondo, E. Pohl, Z. Dauter, K.S. Wilson, M. Hervas, J.A. Navarro, M.A. De la Rosa & G.M. Sheldrick (1995). Structure 3, 1159-1169.

R. Miller, G.T. DeTitta, R. Jones, D.A. Langs, C.M. Weeks & H. Hauptman (1993). Science 259, 1430-1433.

G.M. Sheldrick & R.O. Gould (1995). Acta Cryst. B51, 423-431.