HSPH Biostatistics and DFCI Data Science Seminar
Tuesday April 15 at 4:00pm
Dana-Farber Cancer Institute
Center for Life Sciences Building, 11th Floor, Room 11081

Jian Ma, PhD
Ray and Stephanie Lane Professor of Computational Biology
Carnegie Mellon University

The intersection of Al/ML and biomedicine is entering a transformative era, with growing potential to
impact both basic research and translational medicine. Yet, despite remarkable advances in high-
throughput technologies across genomics and cell biology, our understanding of the diverse cell types
in the human body and the underlying principles of intracellular molecular organization and
intercellular spatial interactions remains incomplete. A central challenge lies in developing
computational frameworks that can integrate molecular, cellular, and tissue-level data to advance cell
biology at an unprecedented scale. In this talk, I will present our recent work on machine learning
approaches for regulatory genomics, with a focus on single-cell 3D epigenomics. We introduce methods
that connect different layers of 3D genome architecture and cellular function at single-cell resolution,
including graph- and hypergraph-based models that capture spatial genome organization. I will also
highlight our latest efforts in developing self-supervised learning frameworks to delineate multiscale
cellular interactions within complex tissues, enabling the discovery of previously unrecognized spatially
organized patterns. Together, these Al-driven models provide a foundation for integrative, multiscale
representations of cellular systems, offering new insights into genome structure, gene regulation, and
cell-cell communication. This line of work opens new opportunities toward building cohesive multiscale
cellular models applicable across a broad range of contexts in health and disease.