The Chain: Protein Engineering Podcast Episode: 70 - Views on Machine Learning and AI in the Antibody and Protein Engineering Space
Mar 12, 2025
Andreas Plückthun, a leading biochemist at the University of Zurich, shares insights on the role of AI and machine learning in antibody and protein engineering. He discusses the importance of epitope specificity in drug development and the challenges in accessing and curating high-quality data. The conversation also covers advancements in predictive models and the necessity of experimental validation to enhance therapeutic protein design. Plückthun emphasizes the need for integrating diverse data modalities and improving methodologies to overcome existing hurdles in the field.
AI Snips
Chapters
Transcript
Episode notes
Defining True AI in Antibody Design
- Andreas Plückthun defines true AI in antibody design as de novo design of human antibodies.
- This involves binding to a specific epitope with high affinity and specificity, a goal yet to be achieved.
AI's Promise and Challenges
- While AI shows promise in de novo protein design, it requires a target structure and doesn't guarantee specificity.
- AI struggles to compete with established antibody technologies like library panning and directed evolution.
Importance of Epitope Definition
- Epitope definition is crucial as it impacts drug function, like preventing receptor binding or dimerization.
- AI could potentially specify epitopes from the start, unlike traditional methods requiring extensive testing.