Abstract
Biological processes are complex, multi-scale phenomena displaying extensive heterogeneity across space, structure, and function. Moreover, these events are highly correlated and involve feedback loops across scales, with nuclear transcription being effected by protein concentrations and vice versa, presenting a difficulty in representing these through existing mathematical approaches. In this thesis we use higher-dimensional spatio-structuro-temporal representations to study biological heterogeneity through space, biological function, and time and apply this method to various scenarios of significance to the biological and clinical communities.We begin by deriving a novel spatio-structuro-temporal, partial differential equation framework for the general case of a biological system whose function depends upon dynamics in time, space, surface receptors, binding ligands, and metabolism. In order to simulate solutions for this system, we present a numerical finite difference scheme capable of this and various analytic results connected with this system, in order to clarify the validity of our predictions. In addition to this, we introduce a new theorem establishing the stability of the central differences scheme.Despite major recent clinical advances, cancer incidence continues to rise and resistance to newly synthesised drugs represents a major health issue. To tackle this problem, we begin by investigating the invasion of aggressive breast cancer on the basis of its ability to produce extracellular matrix degrading enzymes, finding that the cancer produced a surgically challenging morphology. Next, we produce a novel structure in which models of cancer resistance can be established and apply this computational model to study genetic and phenotypic modes of resistance and re-sensitisation to targeted therapies (BRAF and MEK inhibitors). We find that both genetic and phenotypic heterogeneity drives resistance but that only the metabolically plastic, phenotypically resistant, tumour cells are capable of manifesting re-sensitisation to these therapies. We finally use a data-driven approach for single-cell RNA-seq analysis and show that spatial dynamics fuel tumour heterogeneity, contributing to resistance to treatment accordingly with the proliferative status of cancer cells.In order to expound this method, we look at two further systems: To investigate a case where cell-ligand interaction is particularly important, we take the scenario in which interferon (IFN) is produced upon infection of the cell by a virus and ask why biological systems evolve and retain multiple different affinities of IFN. We find that low affinity IFN molecules are more capable of propagating through space; high affinity molecules are capable of sustaining the signal locally; and that the addition of low affinity ligands to a system with only medium or high affinity ligands can lead to a ~23% decrease in viral load. Next, we explore the non-spatial, structuro-temporal context of male elaboration sexual and natural selection in Darwinian evolution. We find that biological systems will conserve sexually selected traits even in the event where this leads to an overall population decrease, contrary to natural selection.Finally, we introduce two further modelling techniques: To increase the dimensionality of our approach, we develop a pseudo-spectral Chebyshev polynomial-based approach and apply this to the same scenario of phenotypic drug resistance as above. Next, to deal with one scenario in which proliferative and invasive cancer cells are co-injected, inducing invasive behaviours in the proliferative cells, we develop a novel agent-based, cellular automaton method and associated analytic theorems for generating numerical solutions. We find that this method is capable of reproducing the results of the co-injection experiment and further experiments, wherein cells migrate through artificially produced collagen microtracks.