Weighting protein ensembles with Bayesian statistics and small-angle X-ray scattering data

Abstract

Intrinsically Disordered Proteins (IDPs) are involved in a number of neurodegenerative disorders such as Parkinson's and Alzheimer's diseases. Their disordered nature allows them to sample many different conformations, so their structures must be represented as ensembles. Typically, structural ensembles for IDPs are constructed by generating a set of conformations that yield ensemble averages that agree with pre-existing experimental data. However, as the number of experimental constraints is usually much smaller than the degrees of freedom in the protein, the ensemble construction process is under-determined, meaning there are many different ensembles that agree with a given set of experimental observables. The Variational Bayesian Weighting program uses Bayesian statistics to fit conformational ensembles, and in doing so also quantifies the uncertainty in the underlying ensemble. The present work sought to introduce new functionality to this program, allowing it to use data obtained from Small-Angle X-ray Scattering.