On the Road from Orthosteric to Allosteric Drugs: The ROAD database of allosteric and orthosteric sites and ligands

We present here the Repository for Orthosteric and Allosteric Design (ROAD), a comprehensive database containing sets of allosteric and functional (orthosteric) sites and ligands. The database responds to a pressing demand for high-quality data in the field of allosteric drug design. Applying a set of strict criteria to crystal structures, we systematically identified 6207 and 20,957 distinct allosteric and functional sites, bound by 4139 and 14,570 small-molecule ligands, respectively, in a total of 3882 proteins. This extensive array of proteins allowed us to analyze structures ranging from single-domain proteins to multidomain and oligomeric assemblies, and to establish the typical range of distances between allosteric and functional sites. Moreover, the comparison of allosteric and functional sites and their ligands reveals major distinctions between them and shows how they are related to the distinct modes of action of allosteric and orthosteric drugs. Realizing the potential of machine learning for the prediction of druggable sites and drug design requires large-scale, annotated training data. ROAD provides this foundation and indicates which archetypal characteristics contribute to the successful design of allosteric and orthosteric drugs. We believe that the comprehensive data in ROAD will also pave the way for the development of basic principles and generalizable approaches for the rational design and engineering of allosteric mechanisms in proteins for a range of biomedical applications. The database and code are freely available at https://github.com/WeiVen-Tee/ROAD-database and the interactive application is available at https://huggingface.co/spaces/weiventee/Repository_for_Orthosteric_and_Allosteric_Design.