Use of the occupancy factor in the refinement of solvent molecules in protein crystal structures

Abstract

A model calculation has been performed to determine whether a variable occupancy factor should be used in the refinement of solvent molecules in protein crystal structures. The atomic structure factor of oxygen was modified by a temperature factor of 0 or 50 A 2 and an occupancy factor of 1.0, 0.75, 0.50 or 0.25, and 'observed' structure factors were calculated for this, the 'target' atom, in a cubic unit cell. The structure factors for a 'model' atom were calculated by keeping the occupancy at a fixed value and by modifying the oxygen structure factor by a temperature factor, B,,. A 'best' Bm was selected by minimizing the square of the differences of the target and model structure-factor amplitudes. The agreement between the electron density of the target and best model atoms is good in all cases except when the target-atom temperature factor was 0 A z and the nominal resolution was 1 A. This agreement suggests that, for data limited to a nominal resolution of not better than 1 A, it is not appropriate to vary both occupancy and temperature factor for solvent molecules in protein structure refinements. Stereochemical considerations suggest that a fixed occupancy of less than 1.0 (e.g. 050) is likely to maximize the electron density fit.