Overview

The modern revolution in single-cell technologies and lineage tracing tools has enabled measurements at unprecedented resolution and scale, generating large, high-dimensional datasets that reveal extensive non-genetic variation and "plasticity" driving diverse cellular outcomes. The concepts of "plasticity," "cell state," and "fate" have surged in various biological contexts, from developmental biology to disease progression, yet often lack clear consensus definitions.

​

The concept of "cell states" was first established by Conrad Waddington in 1957 when he described cellular development following an epigenetic landscape shaped by gene expression. Recently, "cell state" has seen liberal use in various contexts, often interchangeably with "cell type" and "cell identity," resulting in confusion and stalled progress towards the proposed creation of “virtual cells” that can bypass the need for or significantly constrain  physical experiments. Moreover, developing robust connections from cell states to eventual fates remains challenging.

​

Robust regulatory control of these cell-fate decisions is essential in development, homeostasis, and regeneration, as well as during dynamic responses to changing contexts, such as immune responses; however, it gets altered in pathological conditions, such as cancer and autoimmune disease. Recent advances in collecting spatiotemporal data about cellular decision-making have created an urgent need for new mathematical and computational approaches. Tools from dynamical systems theory, information theory, and stochastic processes are now being employed to decode the trajectories that cells traverse during transitions, but significant quantitative challenges remain that require novel mathematical frameworks.

​

This workshop aims to bring together theoreticians, experimentalists, statisticians, and computational modelers to formalize a unified framework for understanding plasticity and cell state-fate relationships. We will explore novel classification and continuous paradigms inspired by physical principles governing biological systems, reframing traditional machine learning challenges through the lens of physical constraints imposed by biomolecular signaling. This integrated approach offers unique advantages where traditional computational methods face limitations: handling extremely limited data, addressing inherently noisy biological measurements, and maintaining interpretability within mechanistic models.

​

Participants will engage with both theoretical foundations and practical implementations, gaining insights into how biologically informed computational architectures can yield more robust, efficient, and interpretable systems for analyzing complex biological data and ultimately predicting cell fate decisions.