Ensemble Reweighting Using Cryo-EM Particle Images

Cryo-electronmicroscopy (cryo-EM) has recently become a leadingmethod for obtaining high-resolution structures of biological macromolecules.However, cryo-EM is limited to biomolecular samples with low conformationalheterogeneity, where most conformations can be well-sampled at variousprojection angles. While cryo-EM provides single-molecule data forheterogeneous molecules, most existing reconstruction tools cannotretrieve the ensemble distribution of possible molecular conformationsfrom these data. To overcome these limitations, we build on a previousBayesian approach and develop an ensemble refinement framework thatestimates the ensemble density from a set of cryo-EM particle imagesby reweighting a prior conformational ensemble, e.g., from moleculardynamics simulations or structure prediction tools. Our work providesa general approach to recovering the equilibrium probability densityof the biomolecule directly in conformational space from single-moleculedata. To validate the framework, we study the extraction of statepopulations and free energies for a simple toy model and from syntheticcryo-EM particle images of a simulated protein that explores multiplefolded and unfolded conformations.