Presentation: Therapeutic Antibody Design Using Generative AI

Presented by: Christian Stegmann, PhD, Senior Vice President, Drug Creation, Absci and Amir Shanehsazzadeh, Senior AI Scientist, Absci

Abstract: Artificial intelligence (AI) has the potential to greatly increase the speed, quality and controllability of drug creation. We showcase such potential on two antibody drug creation tasks: de novo design and lead optimization. For the first task, we utilize generative AI models for zero shot design of antibody CDRs. In particular, we screen over 1 million antibody variants designed for binding to human epidermal growth factor receptor 2 (HER2) using our high-throughput wet lab capabilities. Our models successfully design all CDRs in the heavy chain of the antibody and compute likelihoods that are calibrated with binding. We achieve binding rates that are several multiples higher than HCDR3s and HCDR123s randomly sampled from the Observed Antibody Space (OAS). We further characterize 421 AI-designed binders using surface plasmon resonance (SPR), finding three that bind tighter than the therapeutic antibody trastuzumab. The binders are highly diverse, have low sequence identity to known antibodies, and adopt variable structural conformations. For the second task, we show that our high-throughput screening assay can generate quantitative binding affinity scores for hundreds of thousands of antibody variants. After validating these scores with SPR, we use this data to train large language models that accurately predict binding affinities for unseen antibody variants. These models can be used to co-optimize multiple antibody properties, as we demonstrate by designing variants of trastuzumab with up to seven mutations and with over ten-fold increase in binding affinity. Combined, these approaches promise to accelerate and improve antibody drug creation, and may increase the success rates in developing novel antibody and related drug candidates.