Lab-in-the-loop therapeutic antibody design with deep learning

Nathan C. Frey, Isidro Hötzel, Samuel Stanton, Ryan L. Kelly, Robert G. Alberstein, Emily K. Makowski, Karolis Martinkus, Daniel Berenberg, Jack Bevers, Tyler Bryson, Pamela Chan, Alicja Czubaty, Tamica A D'Souza, Henri Dwyer, Anna Dziewulska, J.w. Fairman, Allen Goodman, Jennifer L. Hofmann, Henry H Isaacson, Aya Abdelsalam Ismail, S. Joseph James, Taylor Joren, Simon Kelow, James R. Kiefer, Matthieu Kirchmeyer, Joseph Kleinhenz, James T. Koerber, Julien Lafrance‐Vanasse, Andrew Leaver‐Fay, Jae Hyeon Lee, E. Lee, D. Lee, Wei‐Ching Liang, J. Lin, Sidney Lisanza, Andreas Loukas, Jan Ludwiczak, Sai Pooja Mahajan, Omar Mahmood, Homa MohammadiPeyhani, Santrupti Nerli, Ji Won Park, Jaewoo Park, Stephen Ra, Sarah A. Robinson, Saeed Saremi, Franziska Seeger, Indraneel Sinha, Anna M. Sokòl, Christoph Spiess, Nataša Tagasovska, Ho To, Edward Wagstaff, Amy Wang, Andrew M. Watkins, Blair Wilson, Shuang Wu, Karina Zadorozhny, John C. Marioni, Aviv Regev, Yan Wu, Kyunghyun Cho, Richard Bonneau, Vladimir Gligorijević

Abstract

Therapeutic antibody design is a complex multi-property optimization problem with substantial promise for improvement with the application of machine-learning methods. Towards realizing that promise, we introduce “Lab-in-the-loop,” a new approach that orchestrates state-of-the-art repertoire mining methods, generative machine learning models, multi-task property predictors, active learning ranking and selection, and in vitro experimentation in a semi-autonomous, iterative optimization loop. By automating the design of antibody variants, property prediction, ranking and selection of designs to assay in the lab, and ingestion of in vitro data, we enable an end-to-end approach to developing computationally-informed therapeutic antibody design pipelines. We apply lab-in-the-loop to eleven seed antibodies obtained via animal immunization with four clinically relevant antigen targets: EGFR, IL-6, HER2, and OSM. Over 1,800 unique antibody variants are tested throughout four rounds of iterative optimization identifying 3–100× better binding variants for all targets and 10/11 seeds, with the best binders exceeding 100 pM affinity, demonstrating a process by which end-to-end machine learning can be developed for therapeutic antibody development.

Introduction

Designing a therapeutic antibody is a "multi-property optimization" challenge. An effective antibody must not only bind tightly to its target (affinity) but also be manufacturable (expression), stable (developability), and highly specific to avoid off-target effects. Traditional discovery methods, such as animal immunization or display technologies, are often linear and struggle to balance these competing requirements simultaneously. While machine learning (ML) offers a way to navigate the vast sequence space of proteins, previous efforts have often struggled with "naturalness" or real-world translation. The "Lab-in-the-loop" (LitL) framework addresses these gaps by integrating generative AI with iterative experimental validation.

Methodology

The LitL framework functions as a semi-autonomous "flywheel" that cycles through computational design and experimental testing.

The Lab-in-the-loop (LitL) workflow integrating generative models, property prediction, and iterative in vitro testing.

The pipeline consists of four primary components:

  1. Generative Ensemble: A mix of "unguided" models (e.g., Walk-Jump Sampling, SeqVDM) for broad exploration and "guided" models (e.g., LaMBO-2, DyAb) for targeted optimization.
  2. Discriminative Oracles: A multi-task framework called Cortex, built on the LBSTER protein language model, predicts expression yield, binding affinity, and non-specificity.
  3. Active Learning: The system employs the Noisy Expected Hypervolume Improvement (NEHVI) acquisition function to identify designs on the Pareto frontier—balancing trade-offs between multiple goals.
  4. High-Throughput Wet Lab: Designs are synthesized and tested in 4–6 week cycles, with the resulting data fed back to refine the models.

The system was pre-trained on massive public databases (UniRef50, PDB, OAS) and fine-tuned on internal Genentech datasets. It was evaluated against four clinical targets: EGFR, IL-6, HER2, and OSM.

Results

The LitL approach demonstrated significant improvements across all targets. Over four rounds of optimization, the team tested over 1,800 unique variants.

Experimental validation results showing affinity improvements across design rounds and correlation between predicted and measured binding.

Key findings include:

  • Affinity Gains: 10 out of 11 starting "seed" antibodies saw at least a 3× improvement, with several reaching 10–100× better binding. The best variants achieved sub-nanomolar affinities (~100 pM).
  • Iterative Learning: Model accuracy improved significantly with each round of lab data, highlighting the value of high-quality, task-specific feedback loops.
  • Structural Integrity: Crystal structures of the designs confirmed that the AI-suggested mutations preserved the original binding modes while optimizing the interface through "rational engineering" behaviors, such as rigidifying CDR loops.

Discussion

While LitL represents a major step toward autonomous drug discovery, it currently relies on high-quality "seeds" from traditional methods rather than full de novo design. Future work aims to incorporate more complex properties like in vivo pharmacokinetics and immunogenicity into the loop. By moving toward a fully integrated "discovery factory," the time and cost of developing life-saving antibody therapies could be drastically reduced.

Summary updated

Abstract

Abstract Therapeutic antibody design is a complex multi-property optimization problem with substantial promise for improvement with the application of machine-learning methods. Towards realizing that promise, we introduce “Lab-in-the-loop,” a new approach that orchestrates state-of-the-art repertoire mining methods, generative machine learning models, multi-task property predictors, active learning ranking and selection, and in vitro experimentation in a semi-autonomous, iterative optimization loop. By automating the design of antibody variants, property prediction, ranking and selection of designs to assay in the lab, and ingestion of in vitro data, we enable an end-to-end approach to developing computationally-informed therapeutic antibody design pipelines. We apply lab-in-the-loop to eleven seed antibodies obtained via animal immunization with four clinically relevant antigen targets: EGFR, IL-6, HER2, and OSM. Over 1,800 unique antibody variants are tested throughout four rounds of iterative optimization identifying 3–100× better binding variants for all targets and 10/11 seeds, with the best binders exceeding 100 pM affinity, demonstrating a process by which end-to-end machine learning can be developed for therapeutic antibody development.
BibTeX
@misc{frey2025lab,
  author = {Nathan C. Frey and Isidro Hötzel and Samuel Stanton and Ryan L. Kelly and Robert G. Alberstein and Emily K. Makowski and Karolis Martinkus and Daniel Berenberg and Jack Bevers and Tyler Bryson and Pamela Chan and Alicja Czubaty and Tamica A D'Souza and Henri Dwyer and Anna Dziewulska and J.W. Fairman and Allen Goodman and Jennifer L. Hofmann and Henry H Isaacson and Aya Abdelsalam Ismail and S. Joseph James and Taylor Joren and Simon Kelow and James R. Kiefer and Matthieu Kirchmeyer and Joseph Kleinhenz and James T. Koerber and Julien Lafrance‐Vanasse and Andrew Leaver‐Fay and Jae Hyeon Lee and E. Lee and D. Lee and Wei‐Ching Liang and J. Lin and Sidney Lisanza and Andreas Loukas and Jan Ludwiczak and Sai Pooja Mahajan and Omar Mahmood and Homa MohammadiPeyhani and Santrupti Nerli and Ji Won Park and Jaewoo Park and Stephen Ra and Sarah A. Robinson and Saeed Saremi and Franziska Seeger and Indraneel Sinha and Anna M. Sokòl and Christoph Spiess and Nataša Tagasovska and Ho To and Edward Wagstaff and Amy Wang and Andrew M. Watkins and Blair Wilson and Shuang Wu and Karina Zadorozhny and John C. Marioni and Aviv Regev and Yan Wu and Kyunghyun Cho and Richard Bonneau and Vladimir Gligorijević},
  title = {{Lab-in-the-loop therapeutic antibody design with deep learning}},
  year = {2025},
  url = {https://openreview.net/forum?id=eQ4tdXPq4N},
}
Figure