Powered by Indico
v3.2.8
"Advances in Deep Learning for CryoEM"
Cryogenic electron microscopy (cryoEM) is a revolutionary but computationally intensive technique for imaging biomolecules in their native state to near-atomic resolution. Mainstream reconstruction pipelines, whose primary task is to infer the unknown 3D density function from thousands of 2D projections, do not use deep learning, but recent work demonstrates the feasibility of integrating neural and existing algorithms. In this journal club, I'll review advances from the last year in deep learning for cryoEM reconstruction. Dependent on time, we will cover some subset of the following three papers: "Reconstructing continuous distributions of 3D protein structure from cryo", "CryoGAN: A New Reconstruction Paradigm for Single-particle Cryo-EM Via Deep Adversarial Learning", and "Exploiting prior knowledge about biological macromolecules in cryo-EM structure determination".