Emprego de GPGPUs para acelerar simulações do sistema humano inato

Abstract

Two mechanisms are used by the Humman Immune System (HIS) to protect the body against diseases caused by distinct pathogens: the innate and the adaptive immune system. The first one is composed of cells and chemicals that use a generic mechanism of defense to prevent or limit infections caused by most pathogens. The second mechanism is activated by the first one. It has the ability to recognize and remember specific pathogens, contributing to the assembly of a more powerful attack each time the same pathogen is encountered again. Despite being widely studied, many questions about the functioning of the HIS are still open because of its complexity and the large number of interactions of its components on distinct levels. In this sense, computational tools are a powerful instrument to assist researchers on this field of study. This work is inserted in this scope and it is split into two parts. In the first part, this work presents the results of a sensitivity analysis on a mathematical-computational model that simulates the innate immune response to lipopolysaccharide (LPS). The main objective of the sensitivity analysis was to find the most sensitive parameters of the mathematical model. The second part of this work proposes the extension of the original model to a three-dimensional one. The simulations in the two parts of the work proved to be computationally expensive, requiring long periods of time to complete. Thus, GPGPUs (General Purpose Graphics Processing Units) were used to reduce execution times. The use of GPGPUs allowed speedups of 276 times for sensitivity analysis, when compared to the sequential one, and of 87 times for computations using the three dimensions model.