LOCATION OF HEAVY ATOMS FROM MAD DATA BY INTEGRATED DIRECT AND PATTERSON METHODS
George M. Sheldrick
Institut für Anorganische Chemie, Universität Göttingen, Tammanstr. 4, D-37077 Göttingen, Germany; gsheldr@shelx.uni-ac.gwdg.de
Keywords: MAD Phasing, Direct Methods, Patterson Superposition
An essential stage in MAD phasing is the location of the anomalous scatterers. If good FA values are available and the number of anomalous scatterers is small, the FA2 Patterson can be interpreted by hand (most educational) or small-molecule direct methods programs can be (mis)used to find the anomalous scatterers. Both of these approaches are less productive when the data are noisy or incomplete, and when there are many anomalous scatters.
Direct methods [1] based on the Shake & Bake philosophy [2] are better at handling incomplete data and can exploit the fact that the number of anomalous scatterers is usually known (though multiple selenomethionine conformations are a complicating factor). A further major improvement [3] is to start the real/reciprocal space recycling procedure from potential atomic sites derived by Patterson superposition rather than from random atoms. In addition, the Patterson superposition minimum function (represented as a 'crossword table' [4]) provides an effective criterion to verify the correctness of each atom located.
MAD phasing depends on small differences in measured intensities and is exquisitely sensitive to systematic errors in the data. However sophisticated the interpretation software is, it is still essential to make every effort to collect the highest quality data possible.
When these methods have been adequately tested - the author has found it rather difficult to persuade people to loan him suitable MAD test data, solved or unsolved - they will be incorporated into the SHELX system, possibly within the current millenium. For the current status of this system, the SHELX homepage should be consulted at: http://shelx.uni-ac.gwdg.de/SHELX/