AWSMelih Yilmaz, Luca Giancardo, Yue Zhao, Edward Lee, Chuanyui Teh, Fangda Xu, Gordon TrangAnnouncement
Monoclonal antibodies are a mainstay of biopharmaceutical development, with over 100 FDA-approved drugs and established manufacturing and regulatory pathways, yet conventional discovery is costly and slow, typically taking six to twelve months from target to lead candidate. AI promises to make antibody development cheaper, faster, and more flexible by designing and characterizing therapeutic antibodies computationally, but scientific questions remain. Development hinges on three factors: the best binding site on the target, which candidates bind most tightly, and whether candidates can survive manufacturing and the clinic. Predictive models for each factor do well on familiar targets and assays but considerably worse on unfamiliar ones, a gap that in-distribution benchmarks have made difficult to measure. Three papers from Amazon Bio Discovery’s science team address research questions about each factor, with methods integrated into the AI-powered application that connects biological AI models with lab services.
Monoclonal antibodies are a key class of biopharmaceuticals with over 100 FDA-approved drugs, but conventional discovery is costly and slow, taking six to twelve months from target to lead candidate. AI-based computational design and characterization promise to make development cheaper and faster, yet predictive models perform well on familiar targets but poorly on unfamiliar ones. In-distribution benchmarks have hidden this gap, making it hard to measure and close. The papers aim to address this by tackling the three core factors: binding site, binding affinity, and manufacturability.
The approach involves three research papers from Amazon Bio Discovery’s science team. Each paper addresses one of the three factors in antibody drug development: identifying the best binding site on the target, predicting which candidates bind most tightly, and assessing manufacturability and clinical viability. The models are designed to improve predictive accuracy on unfamiliar targets and assays, moving beyond in-distribution benchmarks. These methods are integrated into Amazon Bio Discovery, an AI-powered application that connects biological AI models with lab services for testing.
Conventional antibody discovery typically takes six to twelve months from target to lead candidate, with mounting costs and long timelines.
Predictive models for antibody development perform well on familiar targets and assays but considerably worse on unfamiliar ones, a gap hidden by in-distribution benchmarks.
Amazon Bio Discovery integrates biological AI models with lab services, allowing scientists to design and test novel drug candidates computationally.
The three factors are: the best binding site on the target, which candidates bind most tightly, and whether candidates can survive manufacturing and the clinic. Each is addressed by a separate paper from the science team.
The models perform well on familiar targets and assays but considerably worse on unfamiliar ones. This gap is difficult to measure because benchmarks are built around in-distribution accuracy, which doesn't reflect real-world generalization.
Summaries are generated from the original by a language model. Check the original before relying on it.
See the idea
Antibody discovery is a three-stage prediction pipeline, and AI models that excel on familiar targets degrade sharply on unfamiliar ones. This explainer maps where that gap appears and what it costs.
Keep reading
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
One useful email. Unsubscribe anytime.
