Language models trained on evolutionary sequences develop representations that organize protein biology across scales, from atomic structure to function and evolutionary relationships. ESMFold2 translates these representations into fast, accurate atomic-resolution predictions of biomolecular structures and interactions, including antibody-antigen complexes. Coupling ESMFold2 with computational search enables the discovery of miniprotein and single-chain antibody binders across multiple targets, achieving high experimental success rates, nanomolar affinities, epitope specificity, and functional activity. Together, these results show how language models can serve as world models of protein biology, enabling virtual experiments at scale to accelerate biologic discovery and therapeutic design.

Active learning offers an efficient strategy for improving predictive model performance in biologics design by prioritizing the most informative sequences for training. This talk will discuss how active learning can identify sequences expected to maximize model gain through uncertainty, diversity, and information-based criteria. Key practical considerations for implementing active learning in computational biologics applications will also be addressed, including how this approach can accelerate model refinement and reduce experimental burden.

Computational design and immune-derived antibody discovery are often conducted independently with limited information exchange. Here, we first generated large-scale B-cell sequences and functional data following immunization. Sequence function data were used to train and tune structure-based and protein language models. Our approach efficiently generated antibodies with improved epitope diversity, function, and developability.

• Comparing post training approaches for adapting large foundation models to biologics specific data, tasks, and development objectives
• Designing reinforcement learning strategies, reward functions, and feedback mechanisms that guide models towards improved molecular performance
• Integrating experimental results into iterative learning cycles while addressing data quality, reward bias, model reliability, and scalability​

De novo antibody design models lack sufficient training data to reliably generalize. We demonstrate scalable generation of structural training data for machine learning-driven antibody design by linking in silico designs of antibody-antigen complexes to high-throughput experimental binding validation.

De novo design promises intentional creation of functional therapeutics, yet translation into the clinic remains uncertain. Binding, though necessary, is an insufficient design objective. We discuss that function should instead guide the design process and be operationalized through an integrated design-build-test-learn paradigm. This way, experimental characterization actively informs and refines computational design—effectively transforming experimental readouts into active sources of learning instead of means of validation.

The Chatterjee Lab at the University of Pennsylvania develops new generative models to design functional biologics. Our work has centered on language models that de novo design peptides to bind and modulate undruggable disease targets. To enable therapeutic translation, we have developed discrete diffusion models to generate peptides that are "Pareto-optimal" across key ADMET properties. Recently, we have extended these frameworks further to discrete flow matching models that generate and refine highly specific binders under competing therapeutic objective, and have extended these frameworks to peptide-drug conjugates, isoform, motif, and conformationally selective binders, and molecules capable of directing cellular state transitions,

Physics-based energetic scoring functions provide high-fidelity estimates of antibody-antigen binding but are too computationally expensive for high-throughput screening. We introduce a knowledge distillation framework that trains a lightweight neural model to inform binding affinity by learning from energetic scores generated by structural scoring tools. This approach preserves much of the predictive accuracy of the original physics-based scores while dramatically reducing inference cost. We demonstrate that the distilled model generalizes across diverse antibody-antigen pairs, offering a practical, scalable solution for integrating structure-informed binding prediction into early-stage antibody discovery pipelines.

The specificity of alpha/beta T cell receptors (TCRs) for peptide epitopes underpins adaptive immunity. The ability to predict and design TCR-epitope recognition could open new therapeutic and diagnostic avenues with transformative potential. In this talk, I will describe work toward this goal, leveraging 3D structural modeling and deep neural networks to predict TCR-epitope pairing and to design TCRs and TCR-mimic antibodies.

Biologics R&D is generating increasingly complex data across sequences, structures, assays, developability, and workflows. This talk highlights how integrated biologics data platforms, combined with AI/ML, can improve discovery by connecting fragmented data, enabling predictive antibody design and optimization, supporting earlier developability assessment, and accelerating more confident candidate selection.

Predicting whether and where an antibody binds its antigen still depends on experimental measurement. We show how federated learning across pharma partners' proprietary antibody-antigen data trains co-folding models, a binder-confidence head, and an affinity-ranking capability tied to physically measured binding. We report how this was done operationally—from data harmonization to federated training across partner environments—and aim to share first results.

Five major pharmaceutical companies are collaborating via federated computing to train models on antibody developability, a class of biologics properties hard to characterize, due to relatively low-throughput wet-lab assays. Federation provides access to collective data that no single company has, with better economics and speed than generation of such data. We will share early results and lessons from the consortium, highlighting the potential impact of advancing prediction in this space.