Computational analysis of biological networks: Measuring evolutionary rewiring and predicting regulatory relationship

Biological networks represent various types of molecular organizations in a cell. In the previous decade, large amount of network data have accumulated that facilitates our knowledge of the composition, topological structure, and functional significance of biological systems. Recently, great scientific achievement has been made to unravel inter- and intra- species variations at both molecular and system levels. Understanding how biological networks evolve could eventually help explain the general mechanism of cellular system. To this end, this thesis investigates the evolution of biological networks in terms of network rewiring. It compares rewiring rate differences among the common types of biological networks utilizing experimental data across species. Then it applies the rewiring rate formulism to show that regulatory networks generally evolve faster than non-regulatory collaborative networks, which is consistent among all species compared. It goes on to address network data quality issue and to computationally model the process of network rewiring with a simulation algorithm. Currently, building high quality biological networks is still the main goal in the system biology community. The final part of this thesis introduces a novel approach to predict transcription factor (TF) target genes in yeast, with significantly better prediction power than previously reported methods. It identifies histone sensitive and insensitive TFs to be distinct and biologically meaningful clusters.