Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only! Grab it Explore a cutting-edge approach to protein representation learning in this 30-minute conference talk by Can (Sam) Chen from Valence Labs. Dive into a novel structure-aware protein self-supervised learning method that effectively captures structural information of proteins. Learn about the innovative use of graph neural networks (GNN) to preserve protein structural information through self-supervised tasks, focusing on pairwise residue distance and dihedral angle perspectives. Discover how the integration of protein language models enhances self-supervised learning through a unique pseudo bi-level optimization scheme. Examine experimental results from supervised downstream tasks that demonstrate the effectiveness of this groundbreaking approach in advancing protein modeling and drug discovery.