RUNNING TITLE Equilibrium Sampling with Density Restraints SUMMARY

The derivation of molecular models from spatial density data generated by X-ray crystallography or electron microscopy is an active field of research. Here, we introduce and evaluate an approach relying on the equilibrium sampling of energy landscapes describing restraints to experimental input data. Our procedure combines density restraints with replica exchange methodologies in the parameter space of the restraints, and we demonstrate its applicability to both flexible polymers and the assembly of protein complexes from rigid components. For the most difficult system studied, we highlight the importance of advanced data analysis techniques in mining poorly converged data further. Successful and unbiased interpretation of input density maps is a prerequisite for using this approach as an auxiliary restraint term in molecular simulations. Because these simulations will also utilize physical interaction potentials, we hope that they will contribute to deriving families of structural models for input data that are ambiguous per se. HIGHLIGHTS • B-splines are used to describe restraints to density maps with implicit background • Advanced sampling methods yield correct solutions for 3 examples • Analysis of simulation data as statistical ensembles is possible • Advanced data mining is helpful for poorly converged cases