We use statistical physics and machine learning approaches to investigate sequence-structure-function relations in proteins. A common theme of our research is how interactions give rise to collective phenomena and complex emergent behaviors.  At the level of genes, we are interested in epistasis - the complex entanglement phenomenon that causes amino acids to evolve in a concerted fashion - and how this shapes molecular evolution. At the cellular level, we investigate how intermolecular interactions drive biomolecules toward self-organization and pattern formation. Toward these goals, we apply and actively develop an extensive arsenal of theoretical and computational approaches including statistical (mean)field theories, Monte Carlo and molecular dynamics simulations, statistical inference of generative models, and deep learning.