Uncovering Protein Ensembles: Automated Multiconformer Model Building for X-ray Crystallography and Cryo-EM

Abstract With the advent of AlphaFold, protein structure prediction has attained remarkable accuracy. These achievements resulted from a focus on single static structures. The next frontier in this field involves enhancing our ability to model conformational ensembles, not just the ground states of proteins. Notably, deposited structures result from interpretation of density maps, which are derived from either X-ray crystallography or cryogenic electron microscopy (cryo-EM). These maps represent ensemble averages, reflecting molecules in multiple conformations. Here, we present the latest developments in qFit, an automated computational approach to model protein conformational heterogeneity into density maps. We present algorithmic advancements to qFit, validated by improved Rfree and geometry metrics across a broad and diverse set of proteins. Automated multiconformer modeling holds significant promise for interpreting experimental structural biology data and for generating novel hypotheses linking macromolecular conformational dynamics to function.