Mathematical modeling of heterogeneous stem cell regeneration: from cell division to Waddington's epigenetic landscape

TL;DR

This paper reviews mathematical models of stem cell regeneration, focusing on cell division strategies and their connection to Waddington's epigenetic landscape, integrating gene network dynamics and machine learning insights.

Contribution

It introduces differential-integral equation models for tissue development and demonstrates how machine learning links single-cell data with dynamic biological processes.

Findings

Differential-integral equations effectively model cell population dynamics.

Machine learning can extract low-dimensional epigenetic states from single-cell data.

Models provide insights into tissue development and tumor progression.

Abstract

Stem cell regeneration is a crucial biological process for most self-renewing tissues during the development and maintenance of tissue homeostasis. In developing the mathematical models of stem cell regeneration and tissue development, cell division is the core process connecting different scale biological processes and leading to changes in cell population number and the epigenetic state of cells. This chapter focuses on the primary strategies for modeling cell division in biological systems. The Lagrange coordinate modeling approach considers gene network dynamics within each cell and random changes in cell states and model parameters during cell division. In contrast, the Euler coordinate modeling approach formulates the evolution of cell population numbers with the same epigenetic state via a differential-integral equation. These strategies focus on different scale dynamics, respectively, and result in two methods of modeling Waddington's epigenetic landscape: the Fokker-Planck equation and the differential-integral equation approaches. The differential-integral equation approach formulates the evolution of cell population density based on simple assumptions in cell proliferation, apoptosis, differentiation, and epigenetic state transitions during cell division. Moreover, machine learning methods can establish low-dimensional macroscopic measurements of a cell based on single-cell RNA sequencing data. The low dimensional measurements can quantify the epigenetic state of cells and become connections between static single-cell RNA sequencing data with dynamic equations for tissue development processes. The differential-integral equation presented in this chapter provides a reasonable approach to understanding the complex biological processes of tissue development and tumor progression.