Our group focuses on computational approaches to discover and characterize novel subtypes of complex diseases. Our long-term goal is to advance precision treatment by identifying disease subtypes with distinct biological etiology, clinical presentation, and/or treatment responses. Our subtyping methods build on basic tools from statistical genetics, functional genomics, and linear algebra. We also develop methods for gene-environment interaction, gene-gene interaction, pleiotropy, and confounder correction, which can be invaluable for characterizing subtypes. We apply these methods to disease areas with dramatic and under-characterized heterogeneity. Currently we focus on asthma, psychiatry, and single cell genomics.