Computational Software for Building Biochemical Reaction Network Models with Differential Equations

The cell is a highly ordered and intricate machine within which a wide variety of chemical processes take place. The full scientific understanding of cellular physiology requires accurate mathematical models that depict the temporal dynamics of these chemical processes. Modelers build mathematical models of chemical processes primarily from systems of differential equations. Although developing new biological ideas is more of an art than a science, constructing a mathematical model from a biological idea is largely mechanical and automatable. This dissertation describes the practices and processes that biological modelers use for modeling and simulation. Computational biologists struggle with existing tools for creating models of complex eukaryotic cells. This dissertation develops new processes for biological modeling that make model creation, verification, validation, and testing less of a struggle. This dissertation introduces computational software that automates parts of the biological modeling process, including model building, transformation, execution, analysis, and evaluation. User and methodological requirements heavily affect the suitability of software for biological modeling. This dissertation examines the modeling software in terms of these requirements. Intelligent, automated model evaluation shows a tremendous potential to enable the rapid, repeatable, and costeffective development of accurate models. This dissertation presents a case study that indicates that automated model evaluation can reduce the evaluation time for a budding yeast model from several hours to a few seconds, representing a more than 1000-fold improvement. Although constructing an automated model evaluation procedure requires considerable domain expertise and skill in modeling and simulation, applying an existing automated model evaluation procedure does not. With this automated model evaluation procedure, the computer can then search for and potentially discover models superior to those that the biological modelers developed previously.