Model Abstraction and Temporal Behavior Analysis of Genetic Regulatory Networks
Abstract
With advances in technologies such as high throughput data collection and genome sequencing methods, the Human Genome Project which, among other things, determined the human DNA sequence and identified all the genes in human DNA was completed at least two years ahead of time. As these technologies are becoming more accurate, efficient, and cost effective and a massive amount of genomic and proteomic data are becoming available at a rapid pace, we are now in the position to face the challenge to understand how these genes coupled with environmental stimuli orchestrate the regulation of cell-level behaviors. However, understanding such complex systems is very expensive and is most likely impractical with wet-lab experiments alone as the amount and the complexity of data substantially increase, requiring the integration of computational methods to make the process more efficient. To allow for substantial acceleration in computational analysis, this dissertation develops a model abstraction methodology for biochemical systems which systematically performs various model abstractions to reduce the complexity of computational biochemical models. Our methodology is particularly useful for systems with small molecular counts that require the discrete and stochastic representation and thus demand substantial computational requirements. As a case study, this dissertation illustrates the application of individual abstraction methods to such systems. Furthermore, it demonstrates the application of collective abstraction methods at various accuracy levels to temporal behavior analysis of several genetic regulatory networks. This dissertation shows that analysis time of biologically relevant properties of such genetic regulatory networks can be improved from days of work to minutes of work using our methodology while maintaining reasonable accuracy.
