Speaker: Wai Shing Tang (CCM)
Topic: Extracting conformational heterogeneity from Cryo-EM data
Abstract: Cryo-electron microscopy (cryo-EM) has recently become a leading method for obtaining high-resolution structures of biological macromolecules. While cryo-EM provides single-molecule data for heterogeneous molecules, retrieving heterogeneous conformations and the respective ensemble probabilities from cryo-EM data remains a mathematical challenge. Some recent methods model conformational variability by machine learning algorithms, however, these models lack the physical interpretability for downstream biophysical analysis. In this seminar, I present our work on incorporating physical modeling and Bayesian inference to model conformational heterogeneity in cryo-EM data. I first focus on reweighting a conformational ensemble obtained from structural sampling tools, e.g., molecular dynamics simulations, to estimate the ensemble density of a set of cryo-EM images of a flexible protein. Then, I present our recent efforts in refining the model to improve the computational and statistical performance for systems where the prior ensemble is unknown.