Protein-ligand data generation at scale to support computational hit finding and optimization

The ability to discover and optimize small-molecule ligands by using innovative physics-based, machine learning (ML) and artificial intelligence (AI) approaches would transform and democratize the discovery of new small molecule drugs and related research tools (chemical probes), with important implications for the development of new medicines for rare and pediatric diseases, cancer, and women’s health. The development of these methods is being held back by the lack of high-quality protein-ligand data in the public domain. To fill this gap, we have assembled a multi-sector public-private partnership that will generate high-quality, ML-ready, open protein-ligand data at scale and as a public good, and will engage with the research community to build, benchmark and test new machine learning models for small molecule discovery and optimization. Our ambition is to identify drug starting points for >500 human proteins and transform and democratize small molecule drug discovery.