In plain language: PROTACs are large, flexible molecules that bring a disease protein and an E3 ligase together so the cell destroys the disease protein. Predicting how these three-part complexes fit together is hard. With the data science team at Almirall, we tested physics-based and AI docking methods on 43 experimentally solved complexes and showed which strategies work, including the clear benefit of modelling protein flexibility.
Why it matters to clients: Degrader designs come from workflows tested against real structures, so partners know how far the models can be trusted.
Published in the Journal of Chemical Information and Modeling, the study establishes performance benchmarks for PROTAC docking in self-docking and cross-docking scenarios and offers practical guidelines for using docking in targeted protein degradation (TPD) programs. It also shows that AI-based tools can produce unrealistic poses for systems unlike their training data, underlining the value of physics-based refinement.
Read the full article in Journal of Chemical Information and Modeling: https://doi.org/10.1021/acs.jcim.5c01581
