EpiVax Inc, a provider of immunogenicity assessment and sequence optimisation for biologic therapeutics and vaccines, announced on Wednesday the release of AI-enabled enhancements to ISPRI, its proprietary in silico platform for assessing the immunogenicity risk of biotherapeutics.

The release introduces an updated JanusMatrix model, a refined tolerance-adjusted immunogenicity score, and an enhanced anti-drug antibody (ADA) prediction model designed to provide deeper insight into immunogenicity risk.

EpiVax said it has long pioneered computational immunology through the development of advanced in silico algorithms for evaluating biotherapeutic sequences. Foundational to this work are EpiMatrix and JanusMatrix, which evaluate T cell epitope content and human cross-conservation to characterise sequence-derived immunogenicity risk and immune tolerance. ISPRI brings these tools together with other extensively validated methods to assess immunogenicity risk across candidate and impurity sequences.

JanusMatrix 2.1 applies AI to improve the prediction of immune tolerance by weighting human epitope cross-conservation based on expression and prevalence data. According to EpiVax, the result is a more precise characterisation of therapeutic epitopes and their tolerance profiles.

Building on JanusMatrix 2.1, EpiVax has introduced a new immunogenicity score designed to align more closely with clinical immunogenicity outcomes. The JanusMatrix-Adjusted EpiMatrix (JAX) score incorporates JanusMatrix tolerance potential directly into EpiMatrix scores, providing insight into intrinsic effector epitope density and a more biologically-informed assessment of sequence-derived immunogenicity risk across therapeutic modalities.

EpiVax has also updated its ADA prediction model with the release of ADA 2.2. The enhanced model combines epitope-based measures with biophysical attributes and mode of action criteria to provide another assessment of antibody immunogenicity risk, trained specifically leveraging a diverse set of clinical monoclonal antibodies. ADA 2.2 is designed to demonstrate a stronger correlation between predicted and clinically observed immunogenicity, providing greater confidence in risk interpretation alongside the JAX score.