Enhancing MAD F(A) data for substructure determination

Heavy-atom substructure determination is a critical step in phasing an unknown macromolecular structure. Dual-space (Shake-and-Bake) recycling is a very effective procedure for locating the substructure (heavy) atoms using F(A) data estimated from multiple-wavelength anomalous diffraction. However, the estimated F(A) are susceptible to the accumulation of errors in the individual intensity measurements at several wavelengths and from inaccurate estimation of the anomalous atomic scattering corrections f' and f''. In this paper, a new statistical and computational procedure which merges multiple F(A) estimates into an averaged data set is used to further improve the quality of the estimated anomalous amplitudes. The results of 18 Se-atom substructure determinations provide convincing evidence in favor of using such a procedure to locate anomalous scatterers.