Twist Bioscience to Provide Antibody Data Services for Lilly TuneLab

Twist Bioscience Corporation (TWST ) announced on September 16, 2026 an agreement with Lilly TuneLab, a collaborative artificial intelligence and machine learning drug discovery platform created by Eli Lilly and Company (LLY ). Under the agreement, the South San Francisco-based company is a provider of antibody characterization data services for TuneLab, including for AbLab, an antibody developability prediction model.

TuneLab was created to accelerate biotechnology innovation by enabling participating companies to access AI/ML drug discovery models trained on decades of Lilly’s proprietary research data, and participating companies may use the prediction models to more rapidly down-select candidate molecules. Through the Twist agreement, users will be able to order antibody services from Twist using TuneLab preferred Twist protocols to generate wet lab data.

“The quality of an AI model’s output depends on the quality of the data used to train it,” said Emily M. Leproust, Ph.D., CEO and co-founder of Twist Bioscience. Leproust said Twist has the scale and high-throughput capabilities to generate the consistent and reliable antibody characterization data the models need for training. She said TuneLab users can select preferred protocols to generate data with preferred pricing, helping them more efficiently select antibody sequences against their targets and contribute that data back to TuneLab for federated training. The company said the approach connects experimental data with AI-enabled discovery, allowing researchers to potentially accelerate antibody drug discovery.

Lilly TuneLab is part of Lilly Catalyze360, alongside Lilly Ventures, Lilly Gateway Labs, and Lilly ExploR&D, which together support biotech innovation by providing access to strategic capital, lab space and technology, and research and development capabilities.

TuneLab’s Federated Learning Model

Eli Lilly and Company launched Lilly TuneLab on September 9, 2025. Lilly estimates that the first release of AI models on the platform includes proprietary data obtained at a cost of over US$1 billion, and said TuneLab is powered by its full drug disposition, safety, and preclinical datasets representing experimental data obtained with hundreds of thousands of unique molecules. In return for access, selected biotech partners contribute training data, which Lilly said fuels continuous improvement of the platform for others in the ecosystem.

Lilly said the platform is hosted by a third party and employs federated learning, a privacy-preserving approach that enables biotechs to use Lilly’s AI models without directly exposing their proprietary data or Lilly’s. “Lilly TuneLab was created to be an equalizer so that smaller companies can access some of the same AI capabilities used every day by Lilly scientists,” said Daniel Skovronsky, M.D., Ph.D., Lilly’s chief scientific officer and president of Lilly Research Laboratories and Lilly Immunology. Nisha Nanda, Ph.D., group vice president and head of Lilly Catalyze360, said most small biotechs lack access to the large-scale, high-quality data needed to train effective models.

According to the official TuneLab site, the platform’s proprietary foundational AI/ML models are trained on 500,000+ preclinical datapoints collected over more than 20 years, including in vivo and in vitro pharmacokinetic and toxicology data, and have broad applicability for small molecules and antibody-based therapeutics. The site states that TuneLab is hosted by Rhino Federated Computing, a third-party software company whose federated learning platform is built on NVIDIA’s NV FLARE framework. Under the workflow described on the site, a global model is distributed from a central orchestration server to each participating company’s node and trained on local data; only model updates, such as mathematical weights and parameters, are transmitted back to the central server, which aggregates the updates to improve the global model.

The site also sets out data-contribution minimums for membership: small molecule companies are asked to contribute a minimum of 1,000 experimental readouts (defined as one molecule observed in a wet lab experiment to measure a property), while antibody companies are asked to contribute a minimum of at least one measured property or endpoint with 125 sequences.

Twist’s Antibody Production Capacity

On its antibody production and characterization page, Twist states that it produces antibodies with an industrial capacity of 6,000+ IgGs per week, with turnaround times starting at 10 business days for HEK293 expression and 13 business days for CHO expression, and total turnaround of 10 to 15 business days for HEK293 and 13 to 18 business days for CHO. The company lists average 1 mL yields of approximately 180 micrograms for HEK293 and approximately 93 micrograms for CHO, automated lines producing from tens to thousands of antibodies per order, and a starting price of $102 per purified IgG for both cell lines.

Twist states that it delivers characterization data in as few as 15 business days. Its binding assessment services include affinity ranking via Carterra LSA (SPR) or Octet (BLI), epitope binning, and binding confirmation through ELISA or on-cell flow cytometry. Developability profiling covers thermal stability measurement via nanoDSF, aggregation and solubility analysis using aSEC, AC-SINS, and DLS, and polyreactivity and CIC screens. Advanced characterization includes intact mass spectrometry, endotoxin testing, and cross-reactivity assessment across species models.

Twist positions the service for AI/ML model training, stating that it delivers high-fidelity, model-ready datasets at industrial scale to fuel discovery pipelines and that its infrastructure supports orders ranging from a few candidates to thousands of antibodies for model training.