High-level Modeling of Biological Networks

Publisher Summary This chapter presents three different high-level approaches on Partial-least-squares Modeling of Signaling Networks, Bayesian Modeling of Transcriptional Networks, and Agent-based Modeling of Cellular Networks. Each approach is focused on a distinct biological problem, to give a sense of the breadth of questions that can be addressed by high-level modeling. “High-level” models can test biological hypotheses, incorporate prior knowledge, and combine biological processes with less emphasis on the detailed molecular mechanisms. Using high-level models, complex biological networks are simplified to a minimal description that is sufficient to explain and predict the available experimental data. High-level models are useful systems tools for situations in which the important biological questions are defined but essential molecular detail is lacking. In place of differential equations, high-level modeling approaches draw from probability and linear algebra to analyze and predict experimental data. The reliance of high-level models on biological measurements may be undesirable in theory. However, in practice, it allows high-level modeling to be more effective at revealing new relationships between molecules or cells, and in determining whether known pathways are supported by experiment.