Structure-based drug repositioning before and after Alphafold

How can a malaria drug be beneficial in cancer, a cancer drug in auto-immune diseases, or a pain killer in an infectious disease? Despite millions of proteins with different sequence and structure, the diversity of binding sites is limited, so that one drugs does not only bind its intended target, but also many off-targets. While this is detremental in many circumstances, it pave the way to exploit this effect in drug repositioning.In this talk I will show how protein structures can be analysed to systematically search for similarities in binding across large structure databases and how to exploit them to formulate drug repositioning hypotheses. The key to move the focus away from drugs and from target towards their interaction. This can be achieved by characterising drug-target interactions in a representative, standardized fingerprint, which makes any pair of drug-target interactions comparable. Screening all drug-target interactions in the PDB, I will present structural details that explain how a malaria drug can be beneficial in cancer, a cancer drug in auto-immune diseases, and a pain killer in an infectious disease. In all cases, the key are similariities in drug-target interactions across vastly different disease applications. The approach is limited to PDB, a large resource with 100.000 structures of which the majority contain ligands. However, it is dwarved by the recently released Alphafold database with 200.000.000 structures. However, they do not have any ligands, which is required for the screening approach of this talk. I will conclude by discussing the opportunities and challenges of overcoming these limits and making the most out of this new waelth of structural data.