Title:  Towards “Newton’s equations” of cell state transitions --- From statistics to dynamics in single cell data analyses
Abstract:  The scientific revolutions in astronomy and physics during the 16th and 17th centuries, driven by researchers like Brahe, Kepler, and Newton, provide a classic example of how data acquisition and analysis can lead to the formulation of universal dynamical systems theories. Similarly, systems biology aims to leverage dynamical systems theory to uncover the qualitative and quantitative relationships between cellular components and their functions within broader regulatory networks. Successful applications have included deciphering network motifs and whole-cell models in prokaryotes. However, the modeling of mammalian cell dynamics remains a significant challenge. Recent advances in high-throughput techniques, particularly single-cell and spatial genomics, have generated vast, multimodal datasets. A new frontier in the field is integrating these datasets into systems biology models to study complex cellular processes. A key long-term goal is to develop "digital twin" models—from single cells to organs and entire organisms—similar to the molecular dynamics simulations of macromolecules pioneered by Karplus, Levitt, and Warshel. In this talk, I will explore our efforts to derive the governing dynamical equations of cellular state transitions using both snapshot and time-series single-cell data. I will first briefly summarize our previous work, then discuss new approaches that integrate chemical physics, nonequilibrium statistical physics, dynamical systems theory, differential geometry, and machine learning/deep learning. I will specifically present a series of data-driven studies on how aging affects the differentiation-versus-cycling decision in muscle stem cells and attempts on rejuvenating the aged cells.