Maximum entropy calculation of electron density with native and single isomorphous replacement data

Abstract

Maximum entropy is applied to the calculation of electron density maps from native and single isomorphous replacement (SIR) intensity data. Native intensity data alone at around 3 Å, resolution are shown to be an insufficient constraint to give an interpretable map. When the method is applied to SIR data at the same resolution, either by direct selection between the 'most probable' phases, or by using both intensity data sets directly as constraints, the result is a significant improvement over a conventional 'best' map, as demonstrated by a calculation on data synthesized from a protein fragment. The robustness of the method is demonstrated by a series of calculations using increasingly noisy data.