FEATURING: 

Dr. Joachim Frank, Columbia University

Cryo-EM data sampling multiple conformations of a molecule are by nature heterogeneous.  Previous approaches to heterogeneity, such as Maximum Likelihood classification, assume a small number of discrete states, and fail in situations where the variability is continuous and encompasses several degrees of freedom.  In collaboration with Abbas Ourmazd, University of Wisconsin at Milwaukee, we have developed a radically different approach based on manifold mapping.  This approach assumes that the variability of projection data resulting from conformational changes is small compared with the variability resulting from changes in viewing direction.  An application to 850,000 cryo-EM  images of the yeast ribosome is shown as a demonstration.

Prof. Joachim Frank is a member of the National Academy of Sciences, a Howard Hughes Medical Institute Investigator, and he was recently awarded the Benjamin Franklin Medal in Life Science.