In the past decade, technologies such as the DNA microarray and ChIP-on-chip have generated a large amount of high-throughput data for biologists. Although these data has provided us systems-level information about gene regulation, a major challenge in systems biology is to derive methodologies that will infer the underlying dynamics and mechanisms of gene regulation. This thesis research is focused on understanding these mechanisms of transcriptional regulation using systems biology approaches. Transcription regulatory networks play an important role in mediating external stimuli and coordinating responses to changing environments. Different methods that infer regulatory interactions directly from microarray data have been developed in the recent past. However, the implicit assumption in these methods that the transcription factor (TF) mRNA expression can be used as a proxy of its activity at protein level is not always correct, due to post-transcriptional and post-translational modifications of TFs.

In this study, a method named iARACNe was developed. It uses the inferred TF activities to estimate the regulatory activity between TFs and their targets. The study demonstrated that the accuracy of the inferred networks using this method was greatly improved. Two additional methods, OmniMiner and coEDGi, which allow a better understanding of the physical interactions between TFs and target genes, were developed in this thesis research. OmniMiner detects and predicts the potential binding sites for the TFs of interest, while coEDGi enables identification of common enhancers upstream of co-regulated genes. Compared to other approaches which only allow isolated analyses, the systems biology approaches developed in this research provide an opportunity for biologists to study transcriptional regulations from both functional genomics and regulatory sequence perspectives simultaneously.