Current Research

Machine Learning-Guided Engineering of Multifunctional Proteins

I apply methods in machine learning to guide the simultaneous optimization of multiple protein functions. This process involves training neural networks on large experimentally collected datasets detailing the functions of many protein variants. Because I am interested in engineering proteins with novel specificities, I employ multi-task machine learning to teach these networks an approximation of the multifunctional sequence-fitness landscape for these proteins. We can leverage these learned understandings to engineer new proteins that occupy functionally specific niches in the fitness landscape. I have applied these methods to engineer a bacteriophage protein to allow phage to target different strains of bacteria specifically and will characterize the function of these engineered proteins in the lab.