Molecular replacement with NMR models using distance-derived pseudo B factors

The statistical significance of molecular-replacement solutions with models derived from NMR data is strongly enhanced if pseudo B factors that reflect expected atomic coordinate errors are introduced. These B factors are derived from atomic distances of an ensemble of NMR models and an averaged model. A recently determined X-ray structure of a Pleckstrin homology domain:ligand complex has been used as a test case for molecular replacement with NMR templates. The feasibility of the models for molecular replacement has been studied in two steps: (i) correctness of solutions, verified by correct rotational angles and translations; (ii) statistical significance (detection) of solutions, measured by R factors and correlation coefficients in the translation function. None of the models with uniform B factors were detectable in the translation function whereas two models with distance-derived B factors gave statistically significant R factors and correlation coefficients. The potential impact of distance-derived B factors on the detectability of molecular-replacement solutions was further tested by using B factors that were derived from spatial distances between each NMR model and the X-ray structure. It is concluded that the introduction of distance-derived B factors can be an essential component of many future molecular-replacement problems that use NMR models as templates.