Protein Interaction Fingerprinting with Geometric Deep Learning

This lecture by the instructor focuses on deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning. It covers topics such as protein structure and function, surface fingerprint-driven protein design, and the challenges in designing protein-protein interactions computationally. The lecture delves into the MaSIF framework for generating fingerprint descriptors, the MoNet architecture for deep learning on non-Euclidean domains, and the end-to-end learning approach for protein surface prediction. It also discusses the performance of MaSIF in predicting surface chemical features and the future perspectives of deciphering fingerprints on interactomes. The lecture concludes with insights on de novo design of protein interactions and the distinctive points in the modeling framework.