Density Modification in X-Ray Crystallography

Crystallographic models are built by interpretation of an experimental image, the electron density map. This map is generally calculated from amplitudes measured experimentally and phases obtained with the multiple isomorphous replacement method. This method has poor precision, generating errors in the phases and therefore in the map. If the quality of the map is not sufficient to trace clearly a molecular model, it is necessary to improve the phases in order to obtain an interpretable map. Density modification methods achieve this by the application of physically meaningful constraints in real space, such as positivity, boundedness, electron density histograms, atomicity at high resolution, uniformity of solvent regions, continuity of the bio-polymer chain, and known noncrystallographic symmetry of the density distribution. To impose the physical constraints on an experimental map, an iterative algorithm has been proposed (1 ,2 ). It alternates real and reciprocal space operations, and merges gradually the physical constraints with the initial amplitudes and phases.