Guilhem Richard is the Chief Technology Officer of EpiVax. He has over 15 years of experience in computational immunology and holds a Ph.D. in Bioinformatics from Boston University. Since joining EpiVax in 2014, Dr. Richard has developed new computational tools to enhance the immunogenicity assessment of novel therapeutic molecules, and to improve the design of new vaccines for infectious diseases and cancer. Appointed CTO in 2025, he now oversees the expansion and modernization of EpiVax’s immunoinformatics technology stack.
Immunogenicity remains a challenge when developing novel biotherapeutics. In silico approaches primarily evaluate immunogenic potentials by predicting HLA ligands. However, not all predicted T cell epitopes result in immune activation, requiring additional screenings to assess immune tolerance and refine immunogenicity analyses.
We introduce here an update to our JanusMatrix algorithm, first designed to assess cross-conservation between therapeutic-derived epitopes and epitopes from the human proteome. JanusMatrix 2.1 enhances assessment of tolerance by integrating expression and prevalence information into a new ML-based model. This new analysis enables the adjustment of immunogenic potentials based on total foreign – non-tolerated – epitope content.
JanusMatrix-adjusted immunogenicity scores provide a better alignment with clinical observations. These improvements were integrated into a new model predicting anti-drug antibody (ADA) responses, leading to a 7-fold increase in the correlation between predicted and observed ADAs, and with over 75% of predicted ADAs falling within 10% of observed values.