Functional inference and pattern discovery from integrated Saccharomyces cerevisiae networks
My thesis focuses on functional inference and pattern discovery from integrated S. cerevisiae networks. The rapid advancement of genomic and proteomic research in the budding yeast has provided us, on a large scale, with information on various characteristics of genes or proteins as well as various types of interactions or relationships between genes or proteins. To represent all above information in a combined manner, we constructed an integrated S. cerevisiae network, in which nodes are genes or proteins, labeled with their characteristics, and links are interactions or relationships between genes or proteins, colored according to their relationship types.
Features of the integrated S. cerevisiae network can be utilized to make functional inferences. We first integrated multiple gene- or protein-pair characteristics using probabilistic decision trees and accurately predicted co-complexed relationship, which may represent functional association between proteins. Next, we demonstrated that direct function prediction can be made successfully by integrating both gene characteristics and various types of gene- or protein-pair relationships using a fully integrated probabilistic model.
The integrated S. cerevisiae network also exhibits recurring interconnection patterns such as network motifs. We explored an integrated S. cerevisiae network composed of five different link types and discovered multi-colored network motifs, as well as network themes—classes of higher-order recurring interconnection patterns that encompass multiple occurrences of network motifs. We also represented the integrated S. cerevisiae network in terms of network themes, which provides a useful simplification of the otherwise confusing tangle of biological relationships.