Deep Learning for Biological Sequence Design towards Personalized Immunotherapy

Abstract: Modern healthcare equipped with scientific studies and large-scale biomedical data has become a data-driven service. Based on personal medical records, physiological data, and personal genomics data, precision healthcare in the modern era provides patients with personalized diagnosis, personalized medicine, and personalized treatment strategies. In this talk, first I will introduce data-driven personalized immunotherapy with biological sequence design. Then I will talk about two different approaches to T-cell receptor sequence optimization and generation, specifically, based on deep reinforcement learning and disentangled Wasserstein autoencoder.

Bio: Martin Renqiang Min is currently the head of the Machine Learning Department, NEC Laboratories America. He received his MSc and PhD degrees in Computer Science from Machine Learning Group, Department of Computer Science, University of Toronto. He did a one-year postdoc at Yale University. His research interests include machine learning and biomedical informatics, focusing on deep generative models, representation learning, and omics for precision medicine. His text-to-video research was reported by many news media including Communications of ACM, Science News, and MIT Technology Review several years ago.