Collective Dynamics of Interacting Cell Types

Recent experimental advances have opened up entirely new ways to study and conceptualize biological systems. This shift is particularly pronounced in the context of complex multicellular organisms, where single-cell technologies allow us to dissect tissue heterogeneity at the level of individual cells. For the theoretical sciences, this wealth of data provides a unique opportunity to synthesize general principles that unify the variability observed across diverse organisms. This thesis focuses on questions that are inspired by the concept of “cell type”, which is fundamental to our understanding of multicellular development and homeostasis, yet remains poorly defined. Motivated by the ongoing data revolution, cell types are defined here as subsets of gene expression space that are invariant to the dynamics of gene regulation. In particular, I investigate physics-inspired dynamical systems that encode cell types as fixed point attractors. In the process, I derive a formal correspondence between Hopfield networks and restricted Boltzmann machines – two classical architectures at the interface of statistical physics and machine learning. I then show that in the presence of gene regulatory noise alone, Hopfield networks encoding mammalian cell types can recapitulate in vivo differentiation patterns. The statistics of noise-induced cell type transitions reflect their transcriptional proximity, suggesting organizing principles for cell lineage trees. Next, by coupling the transcriptional states of multistable single cells, I develop a general model of multicellular gene regulation that formalizes emerging notions of cell type “plasticity” and establishes a framework to study the self-assembly of compositionally diverse tissues from identical subunits. I then analyze the population dynamics of interacting cell types in the context of cancer development, identifying qualitatively distinct collective states and quantifying their roles in cancer initiation. The results of this thesis not only suggest general principles involved in multicellular development and homeostasis, they also point towards new theoretical approaches inspired by multicellular systems.

Department dc:contributor.department

Physics

Year dc:date.issued

2022

Author dc:creator

• Smart, Matthew

Advisor dc:contributor.advisor

• Zilman, Anton

Statement dc:rights

• Attribution-NoDerivatives 4.0 International

Licence dc:rights.uri

• http://creativecommons.org/licenses/by-nd/4.0/

Handle dc:identifier.uri

http://hdl.handle.net/1807/125304

OAI identifier oai:identifier

oai:utoronto.scholaris.ca:1807/125304