Key points are not available for this paper at this time.
Abstract Models describing thermal motion and conformational variability in protein crystal structures were applied to the refinement of a 1.8 Å crystal structure of penicillopepsin. Three methods were tested: conventional refinement using restrained B factors, multiple‐conformer refinement, and time‐averaging refinement using molecular dynamics. The information content of the models was assessed by cross‐validation and by estimating the phase accuracy of the model using phases obtained by multiple isomorphous replacement. The R value always decreased when using multi‐conformer and time‐averaging methods, as compared to conventional refinement. In contrast, the cross‐validated (“free”) R value and the phase accuracy worsened for time‐averaging in vacuum. Inclusion of solvent produced a slight improvement of both measures compared to conventional refinement. Multi‐conformer refinements always improved both measures. An optimum was reached for simultaneous refinement of between four and eight conformers. At the resolution limit of the penicillopepsin data, multi‐conformer refinement is an efficient method to describe conformational variability and thermal motion.
Burling et al. (Sat,) studied this question.