Summary

AI-enabled models offer a promise to explore a far wider set of candidates that is available to experimental methods. Nonetheless, computational methods do not yet translate to experimental outcomes with high-fidelity, especially not being able to generalize.

Join Dr. Konrad Krawczyk (CSO, NaturalAntibody) and Dr. Gordon McInroy (Co-founder, Nuclera) to see how computational prioritization and rapid physical validation can be connected. Konrad will explain how models across sequence, structure, binding, developability, and immunogenicity prioritize candidates and where data gaps limit confidence. Gordon will present a NaturalAntibody-designed anti-PD-1 panel using cell-free full-length IgG expression, ELISA, and SPR, including comparison with CHO reference data. Together, they will show how positive and negative results can tighten the Design-Build-Test-Learn loop and reserve mammalian work for better-supported candidates.

Join us to see how computational design and rapid experimental validation work together to accelerate antibody discovery pipelines.

In this webinar, you’ll learn how to:

• Where computational prediction can be trusted, and where it can't yet. How modeling across sequence, structure, and developability prioritizes candidates and flags liabilities early, and how to tell where public data supports a prediction versus where it runs thin.
• Why de novo designs fail before they ever bind. The practical failure modes of computational design, starting with expression, the first hurdle many candidates never clear, and why catching them early protects downstream time and budget.
• How to turn validation into better models. Why returning both positive and negative results, not just the winners, feeds the next design round and closes the DBTL loop.
• What it takes to confirm binding fast enough to train models. How cell-free protein synthesis screens an AI-generated library for expression, binding, and kinetic ranking before any mammalian commitment.
• Whether crude cell-free data can stand in for CHO. How closely results from unpurified cell-free material track a purified CHO reference on a real anti-PD-1 panel.

Who should attend:

• AI/ML and computational antibody design teams looking to connect model predictions with rapid, decision-ready expression and binding data.
• Heads and directors of antibody discovery, protein engineering, and biologics AI deciding which candidates should progress into mammalian validation.
• R&D leaders and academic researchers building lab-in-the-loop workflows seeking positive and negative experimental data to refine models and accelerate the next design cycle.

Speaker: