Biomolecular condensates
Sequence-resolved simulation of how proteins demix into membraneless compartments, and what tips a liquid droplet towards a solid.
Research
Our work spans method development and application. Each direction below is an active project with students and collaborators attached.
Coarse-grained model development
Deriving coarse-grained interactions bottom-up from atomistic reference data, and mapping the conditions under which they stop being predictive.
Machine learning for molecular models
Learning representations that carry physics rather than hide it, and using learning to find out which parts of a classical force field are actually wrong.