Computational analysis of regulatory networks using genome-scale data

Living cells are the product of gene expression programs involving regulated transcription of thousands of genes. In single-celled organisms such as S. Cerevisiae regulatory networks respond to the external environment, optimizing the cell at a given time for survival in this environment. Thus a yeast cell, finding itself in a sugar solution, will turn on genes to make enzymes that process the sugar to alcohol. Central dogma of molecular biology describes a key assumption of molecular biology, namely, that each gene in the DNA molecule carries the information needed to construct one protein, which, acting as an enzyme, controls one chemical reaction in the cell. Any step during this information flow can be modulated, from the DNA-RNA transcription step to post-translational modification of a protein. The main stages where gene expression is regulated are chromatin domains, Transcription, Posttranscriptional modification, RNA transport, Translation, mRNA degradation and Post-translational modifications. Understanding how the network is manipulated at these different regulatory levels in a co-ordinated way is a major challenge for molecular biologists. Due to the advent of high throughput technology, it has now become feasible to study regulatory networks at the genomic level. Thus the main aim of this thesis is to characterize regulatory networks using genome-scale data. The first step in gene regulation is transcriptional regulation. The association of transcription factors with genes across a genome can be described as a transcriptional regulatory network. Transcriptional networks is widely studied mainly because of availability of data. The expression data publicly available is growing exponentially and is also becoming available for a variety of species. Analyzing these genome-wide expression datasets to build reliable regulatory networks still remains a challenge due to two main problems. Firstly the noisy nature of the data makes it hard to distinguish signal from noise and secondly it provides only a static view of the dynamic system. Despite these disadvantages it is becoming popular amongst experimentalists as it provides a genome-wide view of the cellular system and secondly such experiments are increasingly becoming cost effective. Thus we focus mainly to develop methods inferring transcriptional regulatory networks. Despite the large bias towards characterizing transcriptional regulation, regulatory programs at other regulatory levels are also b