Extending the limits of molecular replacement through combined simulated annealing and maximum-likelihood refinement

Phases determined by the molecular-replacement method often suffer from mod el bias. In extreme cases, the refinement of the atomic model can stall at high free R values when the resulting electron-density maps provide little indication of how to correct the model, sometimes rendering even a correct solution unusable. Here, it is shown that several recent advances in refine ment methodology allow productive refinement, even in cases where the molec ular-replacement-phased electron-density maps do nor allow manual rebuildin g. In test calculations performed with a series of homologous models of pen icillopepsin using either backbone atoms, or backbone atoms plus conserved core residues. model bias is reduced and refinement can proceed efficiently , even if the initial model is far from the correct one. These new methods combine cross-validation, torsion-angle dynamics simulated annealing and ma ximum-likelihood target functions. It is also shown that the free R value i s an excellent indicator of model quality after refinement, potentially dis criminating between correct and incorrect molecular-replacement solutions. The use of phase information, even in the form of bimodal single-isomorphou s-replacement phase distributions, greatly improves the radius of convergen ce of refinement and hence the quality of the electron-density maps, furthe r extending the limits of molecular replacement.