Local scaling: a method to reduce systematic errors in isomorphous replacement and anomalous scattering measurements

Systematic errors between sets of isomorphous replacement and anomalous scattering data may in some cases be reduced by a technique of 'local scaling'. Methods are proposed which may be used to analyse a data set for significant systematic errors, and to determine whether or not 'local scaling' will be worth applying in a given situation. The proposed method also predicts the optimum number of reflections to be included in the determination of the local scale factors. The technique is illustrated by several examples including 'good' and 'bad' isomorphous-replacement data for α-chymotrypsin, and anomalous scattering data for the lysozyme from bacteriophage T4, and for cytochrome b₅₆₂ from Escherichia coli.