Search for a fragment of known geometry by integrated Patterson and direct methods

A method is presented that attempts to exploit all the a priori available information in order to locate a fragment of known geometry in the unit cell. Whereas the orientation of the search model is determined by a conventional but highly automated real-space Patterson rotation search, its position in the cell is found by maximizing the weighted sum of the cosines of a small number of strong translation-sensitive triple-phase invariants, starting from random positions. A Patterson minimum function based on intermolecular vectors is calculated only for those solutions that do not give rise to intermolecular contacts shorter than a preset minimum. This procedure avoids the time-consuming refinement in Patterson space and should be especially efficient for large structures. Finally, the best solutions are sorted according to a figure of merit based upon the agreement with the Patterson function, the triple-phase consistency and an R index involving E_obs and E_calc. Tests with about 30 known structures, using search fragments taken from other published structures or from force-field calculations, have indicated that this novel combination of Patterson and direct methods is reliable and widely applicable. A few selected examples demonstrate the power of the computer program PATSEE, which is compatible with SHELX84 and will be distributed together with it. PATSEE is valid and efficient for all space groups and imposes no limits on the number of atoms or data. The orientation search for a single fragment allows one additional degree of torsional freedom, and up to two fragments may be translated simultaneously.