AI for Biologics Header Image

 





AI for Biologics
From Models to Discovery: Closing the Loop between AI, Biology, and Experimentation
May 18, 2027 ALL TIMES EDT
AI is rapidly changing how biologics are discovered, moving beyond prediction toward the generation, optimization, and design of novel therapeutic molecules. The field is moving toward integrated AI-driven discovery systems that connect multimodal foundation models, generative protein and antibody design, and agentic workflows with experimental biology. The symposium will explore what this next generation of AI-enabled discovery will bring, from designing molecules across multiple therapeutic properties, to using active learning and experimental data to continuously guide the next design cycle. The program will showcase emerging approaches, with a focus on real-world progress, challenges, and opportunities. As AI and experimental science become increasingly connected through a continuous design-build-test-learn loop, the attendees will learn about how these advances can accelerate biologics discovery and bring greater precision to the design of next-generation therapeutics.
Preliminary Agenda

Advancing Biologics Discovery with AI and Automation

ADVANCING BIOLOGICS DISCOVERY WITH AI AND AUTOMATION

Advancing Biologics Discovery with AI and Automation

Photo of Yu Qiu, PhD, Executive Director, Biologics Design and Technology, AstraZeneca , Executive Director , Biologics Design and Technology , AstraZeneca
Yu Qiu, PhD, Executive Director, Biologics Design and Technology, AstraZeneca , Executive Director , Biologics Design and Technology , AstraZeneca

AI is changing how biologics are discovered, designed, and optimized, opening new possibilities for exploring complex molecular design spaces and accelerating the path from concept to candidate. This presentation will examine how AI-driven design and automation can work together to connect computational predictions with experimental evaluation, enabling more iterative discovery, faster learning, and more informed design decisions. The focus is on translating emerging capabilities into tangible advances in biologics discovery.

Designing Beyond Conventional Antibodies

AI-ENABLED DESIGN OF NEW THERAPEUTIC FUNCTIONS AND MULTISPECIFIC BIOLOGICS

Beyond Binding: AI-Guided Design of Antibodies with New Therapeutic Mechanisms

Bowen Jing, Head, Machine Learning, Divergence Labs , Head , Machine Learning , Divergence Labs

Can AI-designed antibodies do more than bind a target? Generative models are opening opportunities to engineer antibodies around specific molecular interactions, geometries, and mechanisms of action. This presentation will explore how AI-guided design can move beyond conventional binding optimization to create antibodies with new therapeutic functions, highlighting an emerging approach to disrupting the IgE–FcεRI interaction and the challenges of translating computational designs into experimentally validated molecules.

One Arm, Two Antigens: AI Design of OR-Gated Multispecific IgGs

Photo of Yanay Ofran, PhD, Founder, CEO, Biolojic Design Ltd. , Founder & CEO , Biolojic Design Ltd
Yanay Ofran, PhD, Founder, CEO, Biolojic Design Ltd. , Founder & CEO , Biolojic Design Ltd

The application of AI to drug discovery focuses mostly on faster and cheaper design of standard antibodies. But to make safer and more efficacious drugs, AI should be used to design antibodies with new properties that existing methods can’t reliably deliver. As an example of one such property, we show how AI models can design antibodies acting as OR-gates while maintaining the developability profile of a standard human IgG. We illustrate the utility of these antibodies with multiple examples, including BD200 and BD9, which are expected to enter the clinic before PEGS Boston 2027.

Developing Better Biologics by Design: AI Powering the Next Leap in Discovery

Photo of Shipra Malhotra, PhD, Senior Scientist, Biologics, Computational Biology and Machine Learning, Takeda , Senior Scientist , Takeda Pharmaceuticals Company
Shipra Malhotra, PhD, Senior Scientist, Biologics, Computational Biology and Machine Learning, Takeda , Senior Scientist , Takeda Pharmaceuticals Company

This talk explores how embedding AI throughout biologics discovery enables better molecules by design. Examples span target prioritization, epitope selection, de novo binder generation, hit triage, lead optimization, and protein redesign. Integrating agentic workflows, generative models, and experimental feedback supports functional selectivity, developability, and manufacturability. The talk highlights practical outcomes, scientific validation, and the evolving role of scientists as orchestrators of continuously learning AI systems across the entire discovery pipeline.

Learning from Data

FROM SEQUENCE TO PERFORMANCE: MAKING AI PREDICTIONS MORE USEFUL IN BIOLOGICS DISCOVERY

What Hundreds of Antibody Campaigns a Year Teach a Language Model

Photo of André A. R. Teixeira, PhD, Senior Director, Antibody Platform, Institution for Protein Innovation , Senior Director , Antibody Platform , Institute for Protein Innovation
André A. R. Teixeira, PhD, Senior Director, Antibody Platform, Institution for Protein Innovation , Senior Director , Antibody Platform , Institute for Protein Innovation

Our end-to-end platform generates antibodies against hundreds of targets each year, for basic research and first-in-class multispecific development. At that scale, each campaign also produces training data. We feed it to language models in DELPHI, an open framework that predicts polyreactivity and aggregation-related behavior from sequence. We test how predictions hold up across antibody libraries, identify the sequence features behind them, and are releasing our trained models for anyone to use.

Bridging the Gap between Prediction and Practice in AI-Guided Antibody Discovery

Photo of Jiayi Cox, PhD, Data Scientist II, Novartis Institutes for BioMedical Research (NIBR) , Data Scientist II , Novartis Institutes for BioMedical Research (NIBR)
Jiayi Cox, PhD, Data Scientist II, Novartis Institutes for BioMedical Research (NIBR) , Data Scientist II , Novartis Institutes for BioMedical Research (NIBR)

Numerous publications have been on model developing in antibody affinity maturation and developability prediction; however, the success is limited when applying on real projects. Here, I will share how we approach data collection, problem definition, model training, results on pipeline, and lessons learned from AI-guided antibody discovery platform.


Integrating AI Across the Biologics Discovery Pipeline

CONNECTING COMPUTATION, EXPERIMENTATION, AND AUTOMATION TO ACCELERATE BIOLOGICS DISCOVERY

Applying AI-Guided Protein Design across the Biologics Pipeline at Sanofi

Photo of Joseph Batchelor, PhD, Head of Structural Biology, US Biologics Research, Sanofi Group , Head of Structural Biology , US Biologics Research , Sanofi
Joseph Batchelor, PhD, Head of Structural Biology, US Biologics Research, Sanofi Group , Head of Structural Biology , US Biologics Research , Sanofi

AI-guided protein design is shifting from proof-of-concept to routine use in biologics discovery. This talk shares how Sanofi combines structural biology, internal and external AI/ML models, and experimental screening to design, evaluate, and de-risk biologics. Using selected case studies from our portfolio, we will discuss what has worked, where current tools fall short, and how tighter integration of computation and experiment is accelerating discovery.

Closing the Loop for Large Molecule: Working at the Speed of Science

Photo of Martina Miteva, Director, Strategic Digital Initiatives, Research & Early Development IT, Bristol Myers Squibb Co. , Dir Strategic Digital Initiatives , Research & Early Development IT , Bristol Myers Squibb Co
Martina Miteva, Director, Strategic Digital Initiatives, Research & Early Development IT, Bristol Myers Squibb Co. , Dir Strategic Digital Initiatives , Research & Early Development IT , Bristol Myers Squibb Co

For large molecules, the DMTA loop has been historically difficult to close. Biologics bring long cycle times, complex expression and purification, multi-parameter developability demands, and data scattered across functions, so promising candidates and ideas wait weeks for the next iteration. The science moves fast, but the cycle often doesn’t.

This talk shares how BMS brought together Computational Sciences, Research, and IT to accelerate the DMTA cycle end to end. We describe the biggest bottlenecks., the technology investments to address the bottlenecks, and the ways of working that turned a series of handoffs into one connected system. Highlights include agent-driven experimental execution, automated data capture, integrated data flow that feeds key information back into design, and the deployment of a democratized Design workspace. 

Since implementation, we have seen a significant reduction in cycle time and greater transparency about program decision-making between teams. Just as important, we share what we learned along the way.

THE FUTURE OF SCIENTIFIC DISCOVERY: HUMAN JUDGMENT AND AI-DRIVEN R&D

PLENARY KEYNOTE PRESENTATION:
When AI Becomes the Scientist's Partner: Human Judgment and the Future of R&D

Photo of Santha Ramakrishnan, PhD, Vice President, Head, R&D Data Strategy and Governance, Bayer Pharmaceuticals , VP, Head, R&D Data Strategy and Governance , Bayer Pharmaceuticals
Santha Ramakrishnan, PhD, Vice President, Head, R&D Data Strategy and Governance, Bayer Pharmaceuticals , VP, Head, R&D Data Strategy and Governance , Bayer Pharmaceuticals
Photo of Anu Sharma, Senior Principal Scientist/Senior Director, MRL BARDS Epidemiology RWE Capabilities and Analytics, Merck & Co., Inc. , Senior Principal Scientist and Senior Director , MRL BARDS Epidemiology RWE Capabilities and Analytics , Merck
Anu Sharma, Senior Principal Scientist/Senior Director, MRL BARDS Epidemiology RWE Capabilities and Analytics, Merck & Co., Inc. , Senior Principal Scientist and Senior Director , MRL BARDS Epidemiology RWE Capabilities and Analytics , Merck

As AI agents evolve from research assistants into active partners in scientific discovery, what becomes the scientist's unique contribution? How can researchers challenge AI-generated conclusions, recognize overlooked opportunities, and preserve scientific judgment? Senior biopharma leaders will explore how human-AI collaboration is transforming R&D decision-making, operating models, workforce capabilities, and accountability, while examining the leadership and organizational strategies needed to harness increasingly autonomous AI systems for scientific innovation and patient impact.

CHANGING THE TRAJECTORY OF DISEASE: FROM PREDICTION TO INTERVENTION

PLENARY KEYNOTE PRESENTATION: Can we Predict Disease before it Happens? Mapping and Redirecting Biological Trajectories

Photo of John Quackenbush, PhD, Professor of Pediatrics and Human Genetics, University of Chicago; Professor Emeritus of Computational Biology and Bioinformatics, Harvard T.H. Chan School of Public Health , Professor of Pediatrics and Human Genetics , University of Chicago
John Quackenbush, PhD, Professor of Pediatrics and Human Genetics, University of Chicago; Professor Emeritus of Computational Biology and Bioinformatics, Harvard T.H. Chan School of Public Health , Professor of Pediatrics and Human Genetics , University of Chicago

Biomedical research has generated increasingly large, rich, and multidimensional datasets, yet we still often study disease by dividing people into discrete categories: healthy or diseased, responder or non-responder, case or control. Biology, however, does not operate in static states. Disease develops along a continuum, shaped by changing regulatory programs and influenced by aging, biological sex, genetics, epigenetics, and environmental exposures.

As our data and computational capabilities advance, an important opportunity is emerging: moving beyond cataloging molecular differences to reconstructing how biological systems change over time. Gene regulatory networks, pseudo-time modeling, and change-point analysis can help identify critical regulatory switches that move a biological system from one state to another and reveal drivers of disease that may be missed when we focus only on individual mutations, genes, or biomarkers.

These approaches are beginning to provide insight into why individuals exposed to similar risks can follow very different disease trajectories, why biological sex can influence disease susceptibility and treatment response, and how aging may reshape the path toward chronic disease. Identifying these transition points may also uncover new therapeutic targets and opportunities to intervene before disease becomes firmly established.

The ultimate vision is a predictive theoretical biology: The ability to understand where an individual is biologically, anticipate where that trajectory may lead, and identify opportunities to redirect it. Moving from describing disease after it occurs toward predicting and ultimately altering its course could transform how we approach target discovery, therapeutic development, and precision medicine.

PLENARY KEYNOTE PRESENTATION: From Data to the Individual: Connecting the Pieces to Deliver Precision Medicine

Photo of Elizabeth Worthey, PhD, Director, Center for Computational Genomics & Data, University of Alabama Birmingham , Director , Ctr for Computational Genomics & Data , Univ Of Alabama Birmingham
Elizabeth Worthey, PhD, Director, Center for Computational Genomics & Data, University of Alabama Birmingham , Director , Ctr for Computational Genomics & Data , Univ Of Alabama Birmingham

Delivering precision medicine requires more than sophisticated technologies or increasingly complex datasets. Understanding individual patients, particularly those with complex diseases and post-infectious syndromes, demands integration of genomic, molecular, clinical, environmental, and longitudinal information. This presentation explores how connecting fragmented expertise, data, and computational capabilities can bridge the gaps between discovery, biological understanding, and clinical action, advancing a more collaborative, integrated approach to individualized diagnosis and treatment.


Register

Conference Tracks

T1: Data Platforms & Storage Infrastructure