Protein structure, dynamics, and assembly through the lens of the computational microscope

Any organism, from a simple bacterium up to a human body, contains millions of proteins with roles as diverse as sensing, transport, defense, motility or structural support. Proteins have characteristic three-dimensional shapes enabling them to interact with designated binding partners (e.g. DNA, drugs or other proteins). These interactions are key for life, making their atomic-level characterization crucial to understand illnesses and inform the design of new therapies. No biomolecule should be considered as a single atomic arrangement though, as biological function often emerges from specific conformational dynamics. Ideally, a full understanding of any protein (mal)function would require accurate knowledge of its whole conformational space in both the absence and presence of its binding partners. Existing experimental techniques cannot typically fully meet this need though. I will discuss molecular modelling and simulations techniques developed by my lab to help address this knowledge gap. I will focus on the common problem of predicting how flexible proteins assemble into complexes, and how machine learning can be used to predict how proteins move.