Stem cell differentiation is governed by complex epigenetic dynamics involving reactive and diffusive biochemical components. This talk presents a novel approach integrating reaction-diffusion theory with machine learning to model these dynamics in hematopoiesis. By coupling reactive-diffusive components, such as transcription factors and mRNA, with purely reactive elements, including cis-regulatory elements, enhancers, and histone modifications, we construct stage-specific reaction-diffusion models for four critical stages of haematopoiesis: embryonic stem cells (ESC), hemangioblasts (HB), hemogenic endothelium (HE), and hematopoietic progenitors (HP). Coefficients for the reaction kinetics of these models are quantified through machine learning applications on experimental datasets, capturing the intricate feedback and regulatory mechanisms underpinning stage-specific epigenetic transitions. Numerical simulations of these models demonstrate high consistency with biologically observed phenomena, providing new insights into the spatial and temporal coordination of epigenetic components. This integrative framework bridges theoretical and experimental biology, offering a new approach for understanding the spatial heterogeneity and stage-specific regulation in stem cell differentiation. The approach not only advances reaction-diffusion modeling in epigenetics but also highlights the potential of machine learning to decipher complex biological systems. These findings pave the way for future applications in regenerative medicine and stem cell engineering, with implications for both fundamental research and therapeutic development.

This talk will be delivered via Zoom; to request the meeting link, email admin(at)iam.ubc.ca.