DIRDIF-96: PATTERSON METHODS, DIRECT METHODS ON DIFFERENCE STRUCTURES, AND COMPLETION OF THE STRUCTURE BY AUTOMATIC RECYCLING. Paul T. Beurskens, Randy Israel, Gezina Beurskens, Rene' de Gelder, W.P. Bosman and J.M.M. Smits, Crystallography Laboratory, University of Nijmegen, The Netherlands

Direct methods become more and more powerful every year .... but some structures cannot be solved easily, because of poor data sets?, low resolution data?, pseudo-symmetry?, heavy atoms?, or just bad luck?

The heavy-atom interpretation techniques in DIRDIF are fully automated and possible symmetry problems are solved by the special application of direct methods to difference structure factors.

Vector search methods offer an ideal possibility to use your chemical knowledge. The expected geometry of a molecular fragment may be obtained from structural publications (your own research, the Cambridge Data Base) or by Molecular Modelling. The intramolecular vectors are rotated in Patterson space, and acceptable orientations are positioned, all without user intervention: the programs are more powerful than the unexperienced user!

Finally, the heavy-atom structure, or the resulting structural fragment from the vector search techniques, is expanded to the full structure: again fast and fully automatic, using improved R₂ recycling criteria, and new strategies for rejecting Fourier peaks by checking expected peak heights and molecular geometries. In most cases the structure is solved.

But if the heavy atoms are not so heavy, or if the Patterson allows homometric solutions, or if the molecular fragment is not unique or not completely correct or very small, a large(?) number of tentative structural models may be obtained. User controlled recycling is then possible, but automatic multisolution-recycling is on its way: the program selects the most probable model (using various FOM's), tries to expand it to the complete structure, and if unsuccesful, selects the next probable model.