Having an accurate protein structure is a necessary condition for modern drug discovery because it provides a mechanistic picture of drug-target interactions and guides the path for the rational optimization of efficacy and therapeutic characteristics. However, in many cases, especially for newer modalities such as biologics and proximity inducers (i.e., PROTACs and molecular glues) which often extend beyond the singleprotein paradigm, structural information is lacking and is hard to obtain using existing computational methods. In addition, static structural information is often not sufficient in those cases. Thus, modeling protein dynamics and the formation of multiprotein complexes is recognized as an important but extremely challenging next frontier for computational chemistry as a whole, and for drug design in particular. Here, we would like to present our most recent results in the development of computational methods for modeling protein dynamics and protein-protein interactions, as well as their applications to complex therapeutic problems such as the design of PROTACs and molecular glues. Our results demonstrate the usefulness of our approach for solving challenging problems in biology and therapeutic design.
Originally presented at the Keystone Symposia 2025.