Encoding and Decoding Specificity in Adaptive Immunity by Deep Learning

Sai Reddy, Ph.D.
Associate Professor
Department of Biosystems Science and Engineering 
ETH Zurich

The ability to predict and correspondingly manipulate adaptive immune responses is highly valuable for biotechnology and medicine. To achieve this requires a greater molecular understanding of antigen selection and specificity by adaptive immune cells. In this I will explain how we are using deep learning to identify patterns of antigen-specificity in antibody responses following immunization. Deep neural networks are used to elucidate the antibody sequence space by generating thousands of novel and functional variants in-silico, highlighting how deep learning can be used to encode and decode specificity in adaptive immunity.