Scientific Environment

ur lab is about evenly split between two areas: the intersection of machine learning and molecular modeling, and the intersection of molecular modeling and systems biology. Roughly speaking, the former focuses on method development of biophysically-grounded machine learning models for molecular problems, including the prediction of protein structure and protein-ligand interactions; the latter focuses on studies that leverage the biophysical and molecular perspective (enabled by the first arm) to address basic and translational questions in systems biology. In this area, we focus on signal transduction networks: how they are organized and how their logic is effected (emphasizing the biophysics and structure of signaling proteins), how they have evolved across metazoa, and how they are dysregulated in disease, particularly cancer. We believe that the biophysical and molecular perspectives provide a particularly powerful lens through which to investigate the genotype-phenotype map.

The boundaries between method development and basic science are not firm however. For example, our systems biology work requires the development of dynamical systems modeling techniques that scale to entire proteomes. Conversely, our machine learning models push the boundaries of the core science of machine learning, for example by broadening learning from Euclidean spaces to general manifolds.