Deep Learned Screenings of Peptide Vaccines for Exosome-Based Precision Delivery

Peptide vaccines are the most biologically relevant route to immunotherapy against disease. However, peptide vaccines are challenged at all stages within the pharmaceutical development pipeline. A grand question is the selection of peptide fragment to deliver to an antigen presenting cell or using nanomaterial systems to bypass and activate a T cell directly. Utilization of artificial intelligence is prevalent to select such peptide fragments, yet is rarely shown to have high efficacy in vivo. Selection of peptide fragments that are cognate for specific TCRs is conducted through statistical string analysis of a diseased cell’s immunopeptidome or from existing consortia databases. Newer techniques have focused on integrating statistical string with molecular modelling at the interface between antigen presentation & TCR recognition, however the therapeutic efficacy has only improved slightly. Deep learning is an advanced artificial intelligence technique capable of integrating molecular mechanisms, experimental data, and clinical cognition, altogether, to predict immunostimulation. We will highlight how deep learning is overcoming challenges in peptide vaccine formulation and translation, while presenting a new perspective into developing peptide vaccines by using deep learning to understand precision drug delivery, specifically aimed at enumerating exosome-mediated activation of T cells.