Project in Prairie
My primary research goal is to develop efficient encoders (now sometimes called foundation models) for the structure of proteins and RNA (i.e. their 3D coordinates). This includes developing benchmarks to compare different encoders, exploring different mathematical representations and tailored deep learning architectures as well as pretraining methods.
Once equiped with powerful representations, we can apply them to downstream tasks of biological interest, notably the prediction of interactions. On the protein side, I focus on protein-protein binding, with applications in predicting if animal viruses could infect humans. I also have a strong interest on RNA-small molecule interaction prediction with applications in drug design, notably for currently uncurable diseases.