Simulation of biochemical systems using constraint-based methods and complex networks

• Ano de defesa: 2021
• Autor(a) principal: Burke, Paulo Eduardo Pinto
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Biochemical networks; Integração de processos; Modelos de célula completa; Process integration; Redes bioquímicas; Simulação estocástica; Stochastic simulation; Whole-cell models
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/95/95131/tde-23032021-143111/

Computational models of biomolecular systems have been employed to advance knowledge in many research areas. They have particular impact in modern areas such as Synthetic Biology, Bioengineering, and Precision Medicine. Current technologies can determine with high precision and throughput the molecules that compose cells and, to a great extent, the interactions between them. These interactions are usually grouped into cellular processes by their function and many are the available models that can represent and simulate them to a certain level of accuracy. Despite being commonly represented as separated systems, they are in fact interconnected. The simple fact that they may share common molecules makes their dynamics dependent on each other. Given the current high availability of data and computational power, models to represent whole-cells are being considered. However, current approaches to model and simulate cellular processes are challenging to be integrated given the high heterogeneity of methods employed. Thus, a more homogeneous approach to represent and simulate could make easier this integration. In this work, we propose a framework to model cellular processes by means of their underlying biochemical reactions as well as a simulation method that sources from this kind of representation. To investigate the capabilities of the modeling framework, we used the organism Mycoplasma genitalium as a case study aiming at representing all the molecules and interactions known to compose this organism by means of a single biochemical network. Among the results obtained from this model, we have that the obtained topology presents a good agreement with the literature, as well as good accuracy on the prediction of essential genes of the organism by employing cascade failure analysis. Additionally, we investigated the characteristics and capabilities of the so proposed simulation algorithm, called CBSA. It is shown to be able to perform efficiently discrete-stochastic evaluations of the dynamics of large sets of interactions. It is also able to be computed in parallel computing architectures such as GP-GPUs. We illustrate this by simulating several theoretical models as well as a challenging real biochemical system. Despite the advances reported in this work, much remains to be done in order to perform simulations at a whole-cell scale by using the proposed methods. Nevertheless, we point out possible future developments aiming at the ultimate goal of performing simulations of whole-cells.