Ginkgo Bioworks announces a series of new initiatives from its Datapoints offering to accelerate the application of artificial intelligence in biologics drug discovery.
These include a strategic partnership with Apheris to launch the Antibody Developability Consortium and, separately, the AbDev AI Competition.
Together, these efforts aim to position Ginkgo Datapoints as a leader in creating the data infrastructure and collaborative frameworks needed to advance antibody AI.
Tackling foundational challenges in drug development
Drug developers face rising costs and longer timelines. AI is positioned to reduce both but is hindered in part by incomplete or siloed datasets, which limit their effectiveness for training AI.
Although historical datasets remain valuable, the industry now requires higher-quality, fit-for-purpose data to train next-generation models.
Through its advanced lab automation, we believe Ginkgo can now generate such datasets in a fraction of the time — helping to complement legacy program data with new data designed for improving AI applications.
Antibody Developability Consortium
The Antibody Developability Consortium, led by Ginkgo Datapoints in partnership with Apheris, aims to address one of the most significant challenges in biologics development: predicting and optimising antibody properties early in R&D to better ensure downstream clinical and commercial success.
- Ginkgo Datapoints will contribute its AI/ML and high-throughput experimental capabilities, creating purpose-built datasets for improved model training.
- Apheris will provide a federated computing infrastructure that enables members to collaborate securely on sensitive data while maintaining full ownership and control.
“The future of AI in drug discovery depends on creating environments where companies can collaborate without compromising their most valuable data,” said Robin Röhm, CEO and cofounder of Apheris.
“We are successfully doing this with the AI Structural Biology (AISB) consortium, another cross-industry initiative powered by Apheris, and now with this new consortium with Ginkgo, we’re bringing federated learning directly into antibody R&D, making critical datasets usable in ways that were never before possible.”
By combining centralised dataset generation with federated model training, the consortium establishes a new framework for collaboration across the industry.
The approach blends high-quality diverse datasets generated by Ginkgo with secure access to distributed partner datasets through Apheris’ federated computing technology.
AbDev AI competition: establishing standards for the field
In parallel, Ginkgo is launching the first-ever AbDev AI competition. The competition is designed to measure the current state of antibody developability modelling and to create widely accepted standards for performance and evaluation.
By providing a transparent, structured environment for testing models, the competition will help to highlight areas of strength in the field and identify where new methods and datasets are most urgently needed.
The AI competition, hosted on the Hugging Face platform, runs now until early November, when winners will be announced with up to $60,000 in prize values.