Bristol Myers Squibb (BMS), Takeda Pharmaceuticals and Astex Pharmaceuticals are the latest companies to join the AI Structural Biology (AISB) Network, a consortium dedicated to training artificial intelligence (AI) models to accelerate drug discovery and development.
“Through this federated learning initiative, we’re bringing together diverse structural data sets from multiple pharma companies to advance predictive models for small molecule discovery in ways no single organisation could achieve alone,” said Payal Sheth, vice president, Discovery Biotherapeutics & Lead Discovery and Optimisation, BMS. “These efforts strengthen our ability to accelerate the development of transformational therapies for patients.”
The AISB Network is led by industry experts in collaboration with the AlQuraishi Lab at Columbia University, US. Its flagship project, OpenFold3, is an AI model designed to predict protein–ligand interactions, a key part of the drug discovery process. The new consortium members will contribute thousands of validated protein–small molecule structures, helping to establish one of the most comprehensive and diverse training data sets in drug discovery and increase the model’s predictive accuracy, according to the partners.
Data for OpenFold3 is collected through a federated data-sharing framework using a secure computing platform developed by German life sciences company Apheris. This approach enables consortium members – including existing participants AbbVie and Johnson & Johnson – to contribute proprietary data while maintaining confidentiality. The aggregated data sets are then used to train and refine the AI model.
“By contributing data from our small molecule–protein structures to a federated effort, we can help ensure models like OpenFold3 better reflect the challenges medicinal chemists face every day, while keeping proprietary science protected,” said Paul Mortenson, senior director, Computational Chemistry and Informatics, Astex.