Last week, researchers at Texas A&M and UTMB, who are part of the ARPA-H APECx “SPHERICAL” project team, convened for an intensive (but fun!) 1.5-day workshop focused on accelerating the progress in AI-driven antigen design.

At the heart of the discussions was the AI-based antigen design pipeline developed by the Texas A&M team, which has been applied to optimize the design of candidate antigens for Mayaro virus (MAYV) and Rift Valley fever virus (RVFV). This represents an important step toward rapid, data-driven countermeasure development that effectively leverages the latest advances in AI, protein language modeling, and protein structure prediction.

A major milestone highlighted during the workshop was the availability of the first batch of experimental results. UTMB successfully synthesized the antigens designed by the AI pipeline and evaluated them using neutralization assays, providing critical validation data for the computational predictions.

Our team engaged in in-depth discussions analyzing the correlation between the predicted antigen properties and the observed neutralization performance. We expect that these insights will be instrumental in refining the current antigen design pipeline, improving the predictive accuracy of the AI models utilized therein, and strengthening the overall design framework in a closed-loop fashion.

The workshop concluded with forward-looking conversations between team members at the two institutions on enhancing the pipeline, including strategies to improve robustness, incorporate uncertainty quantification, and extend its capabilities toward predicting broadly effective antigens against diverse viral variants.

This collaborative effort underscores the power of integrating AI, experimental validation, and interdisciplinary expertise to accelerate the development of next-generation antigen design strategies.

For more information about the ARPA-H APECx program that is sponsoring the project, please visit: https://arpa-h.gov/explore-funding/programs/apecx