BPS2026 – Parsimonious recovery of cryo-EM ensemble probabilities

Heterogeneity analysis has become a very popular area within cryo-electron microscopy. Many bespoke methods exist for estimating what protein conformations are in an experimental sample and how frequently each of these conformations appear. Here, we use a simple approach—treating the estimation as a non-parametric maximum likelihood problem, provide a quick numerical approximation along with theoretical guarantees, practical guidance for integrating with RELION and CryoSparc tools, molecular dynamics simulation, as well as recent machine-learning-based heterogeneity analysis approaches.