DENSITY MODIFICATION: SUCCESSES AND LIMITATIONS. Kevin Cowtan, Protein Structure Group, Dept. of Chemistry, University of York, Helsington, York Y01 5DD, England

Phase improvement by density modification, despite a number of theoretical flaws, is now an established part of the solution of macro-molecular structures by x-ray diffraction methods. The power of the method in the best cases has been demonstrated repeatedly by the solution of structures from the weakest of phase information, for example from single derivatives or weak anomalous scattering. If there is sufficient redundancy in the data, due to high solvent content, local symmetry or multiple crystal forms, then the method can come close to ab-initio phasing.

This success arises despite a number of fundamental flaws in the underlying protocol of most density modification calculations. In particular, the density modification `model' is usually under-determined with respect to the observed data. This leads to degrees of freedom in the model magnitudes which in turn invalidate the assumptions behind the usual phase combination methods.

Future developments must address these problems. Particular areas under study are the integration of model building and density modification into a semi-automatic process. At a more fundamental level, statistical phasing (maximum entropy) methods may lead to a new formulation with a firmer mathematical foundation.