Applied Math Seminar

---

Inferring Compositional and Conformational Probability Distributions Using Cryo-EM Images

Speaker: Pilar Cossio, CCM, Flatiron Institute

Location: Warren Weaver Hall 1302

Date: Friday, April 17, 2026, 2:30 p.m.

Synopsis:

RNAs are versatile molecules involved in nearly all aspects of gene expression. Many RNAs must adopt complex three-dimensional (3D) structures to function effectively. These structures are highly dynamic, sampling multiple conformations under physiologically relevant conditions. Consequently, the dynamic nature of RNA 3D structures limits traditional techniques, such as X-ray crystallography, to capturing only the most stable structure, which may not represent the full functional ensemble. Recent advancements in cryo-electron microscopy (cryo-EM) have made it a powerful tool for studying macromolecules with conformational flexibility. Analyzing the single-particle cryo-EM images, in principle, enables the study of multiple RNA conformations and their relative populations. However, challenges such as low signal-to-noise ratios and the unknown poses of RNA in each single-particle image complicate this task. To overcome these limits, we applied a novel Bayesian cryo-EM ensemble reweighting framework that integrates computational prior ensembles with raw cryo-EM 2D images to infer conformational and compositional distributions.