VIRTUAL IDeAS Seminar: Ellen Zhong, Massachusetts Institute of Technology

Neural reconstruction of protein structure from cryo-EM images

Cryo-electron microscopy (cryo-EM) is a powerful technique for determining the structure of proteins and other macromolecular complexes at near-atomic resolution. In single-particle cryo-EM, the central problem is to reconstruct the 3D structure of a macromolecule from $10^{4-7}$ noisy and randomly oriented 2D projection images. However, the imaged protein complexes may exhibit structural variability, which complicates reconstruction and is typically addressed using discrete clustering approaches that fail to capture the full range of protein dynamics.

This seminar focuses on methods and challenges in addressing protein structural heterogeneity in single-particle cryo-EM reconstruction. I will introduce cryoDRGN, an algorithm that leverages the representation power of deep neural networks to reconstruct continuous distributions of 3D density maps. This method encodes structures in Fourier space using coordinate-based deep neural networks and trains these networks from unlabeled 2D cryo-EM images by combining exact inference over image orientation with variational inference for structural heterogeneity. I will show ab initio reconstruction of synthetic cryo-EM image data with cryoDRGN and show how different types of heterogeneity are reflected in the learned data manifold. I will then highlight a few vignettes applying cryoDRGN on real datasets to reveal new conformations of macromolecular machines and visualize continuous trajectories of their motion. CryoDRGN is open-source software freely available at http://cryodrgn.csail.mit.edu.

Ellen Zhong is a fourth-year Ph.D. student in the MIT Computer Science and Artificial Intelligence Lab advised by Bonnie Berger and Joey Davis. She is interested in problems at the intersection of computation and biology: her current research focuses on developing neural methods for 3D reconstruction of protein structure from cryo-EM images. Prior to MIT, she was a research associate and scientific programmer at D. E. Shaw Research. She obtained her bachelor’s degree from the University of Virginia, where she worked with Michael Shirts on studying protein folding using Hamiltonian Monte Carlo simulations. She recently co-organized the first workshop on Machine Learning in Structural Biology at NeurIPS and is a recipient of an NSF Graduate Student Research Fellowship.