The UK Department for Science, Innovation, and Technology (DSIT) newly established Sovereign AI unit announced an £8 million investment in the OpenBind Consortium to catalyze a transformation in AI for small molecule drug discovery through an accelerated data generation effort centred at the Diamond Light Source.
This dataset will address the critical gap in pharmaceutical R&D: the lack of high-quality, large-scale datasets linking small molecules to the proteins they bind. These datasets are essential for building next-generation AI/ML models capable of transforming early-stage drug design.
Drug discovery hase been undergoing a revolution since the advent of AlphaFold from the UK-based DeepMind. Since then, a firestorm of rapid innovation has led to numerous AI/ML tools for small molecule drug discovery, including AlphaFold3, RoseTTAFold All-Atom, OpenFold, Chai-1, Boltz-2, NeuralPLexer3, and many others. While these AI-based tools are tremendously powerful, they must be built with critical structural and affinity data to extend to new drug targets and chemistries. OpenBind aims to provide the foundational dataset needed for this revolution to fluorish.
OpenBind will create the largest open experimental dataset of protein-small molecule structures and paired affinities in history, with the effort initially focusing on doubling the number of protein-ligand structures in the Protein Data Bank in under two years.
To do this, OpenBind will deploy automated chemistry, protein design, advanced open source AI/ML models, and high-throughput X-ray crystallography to ultimately generate more than 500,000 protein–ligand complex structures and paired high-quality affinity measurements over 5 years. This would represent a 20-fold increase over all public protein-ligand structure data produced in the last half-century, filling a critical gap in the data ecosystem that has slowed the development and evaluation of modern generative models.
Read the full press release here.