The long-term vision of our research program is to uncover the information encoded in the human genome that give rise to the complexity and heterogeneity of the cell through innovations in computational method development. The overarching theme of our work is focused on developing algorithms that would allow us to build better predictive models for fundamental structure (e.g., sequence composition, 3D genome structure in cell nucleus, and spatial organization of the cells in complex tissues) and cell function. We develop new high-dimensional probabilistic models and new representation learning architectures to address fundamental questions in computational epigenomics, gene regulation, and single-cell biology. The results revealed by these algorithms are expected to provide critical insights into genome function changes in different cellular conditions and during disease development. Our lab is currently leading a UM1 Center project in the NIH 4D Nucleome Consortium.