Question 1

How is TuneLab designed to protect proprietary data when training the AI/ML models that use federated learning?

Accepted Answer

TuneLab federated training is hosted by a 3rd party.  During a training run, a global AI model is distributed from a central orchestration server to each participating company's node. Each node then trains this model on its proprietary data, with the raw data remaining in that node. Only model updates, such as mathematical weights and parameters, not the underlying molecular structures, are transmitted back to the central server. The server aggregates these updates from all participating nodes to improve the global model, which is then redistributed for the next training round. This iterative process continues until the model converges to optimal performance. To confirm data quality during this process, Tunelab reviews certain aggregated statistics and analytics derived from the data, together with information about assay conditions and protocols used to generate it.  This review is designed not to disclose, and does not require access to, any underlying molecular structures, sequences, or individual data points.

Question 2

What is Lilly TuneLab?

Accepted Answer

TuneLab is a first-of-its-kind, collaborative AI-powered drug-discovery and development platform that gives biotech researchers access to powerful AI models trained on decades of Lilly proprietary research data and member data contributions. These are the same models used by Lilly's own scientists every day.

Question 3

What kinds of models and use-cases does Lilly TuneLab support?

Accepted Answer

TuneLab currently supports preclinical drug-discovery programs, from first synthesis through to in vivo testing, with core models for small molecules and antibodies designed to work across different data sources and in federated environments.

Question 4

What training data does a biotech contribute in exchange for access?

Accepted Answer

To become a member of Lilly TuneLab, we are asking small molecule companies to contribute a minimum of 1,000 experimental readouts (an experimental readout is defined as one molecule observed in a wet lab experiment to measure a property). For antibody companies, we are asking a minimum of at least 1 measured property/endpoint across 125 sequences. A number of member companies contribute larger amounts of data than these minimums, with the aim of improving TuneLab’s AI/ML models further, and potentially increasing their performance in that specific member’s projects.

Question 5

Who is Lilly TuneLab designed for?

Accepted Answer

Biotechs with at least one preclinical small molecule or antibody program. Every molecule that advances too far before its liabilities costs time, capital, and can delay the development of new medicines for patients. Lilly TuneLab supports predictive characterization of antibody developability as well as small molecule ADMET profiles prior to in vitro or in vivo validation, with the aim that only the most viable candidates progress, and minimizing dead-ends in drug development.

Stay Up to Date with Lilly TuneLab

Grünenthal has joined Lilly TuneLab, a collaborative Artificial Intelligence (AI) and Machine-Learning (ML) platform that provides access to predictive models trained on Lilly proprietary data and strengthens data-driven decision-making in early drug discovery.

Ginkgo Bioworks today announced that its Ginkgo Datapoints offering has entered into a new agreement with Lilly TuneLab, a collaborative AI/ML drug discovery platform created by Eli Lilly and Company.

Twist Bioscience Corporation, a mid-cap growth and value biotech company, today announced an agreement with Lilly TuneLab, a collaborative AI/ML drug discovery platform created by Eli Lilly and Company (Lilly).

Chai Discovery, the AI company engineering models to discover new molecules, today announced that certain Chai Discovery frontier AI model capabilities will be made available to companies participating in the Lilly TuneLab platform.

Charles River Laboratories International, Inc. (NYSE: CRL) announced a collaboration with Lilly TuneLab, a collaborative AI/ML drug discovery platform created by Eli Lilly and Company (Lilly).

Axcelead Drug Discovery Partners, Inc. (Axcelead DDP) is pleased to announce that it has entered a collaboration with Lilly TuneLab, a collaborative AI/ML drug discovery platform created by Eli Lilly and Company (Lilly).

Tamarind Bio has been selected by Lilly TuneLab as a third-party technology partner to build, host, and operate the inference infrastructure layer for TuneLab2.0 which is now live.

Collaborative Drug Discovery (CDD), the leading provider of data management solutions for the pharmaceutical and biotechnology industries, today announced an agreement with Eli Lilly and Company (Lilly) to integrate Lilly TuneLab, a collaborative AI/ML drug discovery platform, into CDD Vault.

AI and machine learning models have the potential to transform drug discovery. Improving access and pooling training data is needed to translate this promise into tangible results.

Pfizer, Eli Lilly, Novartis, Bristol Myers Squibb and AstraZeneca are all ramping up the use of AI, but drug discovery is not the primary success story—yet.