Construct Gene Networks From Different Data

Recently, construction and analysis of gene regulatory networks, which can intuitively represent the causality or regulatory relationships between genes, is veryimportant field in Bio-mathematics. Diverse methods have been offered to set up generegulatory networks from different data to mine the interactions between genes. Inthis paper, two methods are used to construct gene networks from two different data,and new conclusions are obtained through analysis of these networks.The main contents of this paper are as follows.Section one. Construct undirected gene regulatory networks based on mutualinformation theory. The Wilcoxon rank-sum test method is used on gene expressionprofiles in kidney tissues with and without cancer to obtain candidate genes. Thereby,mutual information networks of these genes are constructed. Twenty two structuralkey genes are selected based on the seven statistics of the two networks, and thesegenes are predicted as potential pathogenic key genes of kidney cancer. Empiricalstudies on cancer show that ten of these genes are closely related to the formation anddevelopment of kidney cancer. Furthermore, five pathways are predicted that theymay play an important role of development of kidney cancer based on GO annotation,and three of them have been confirmed.Section two. A new method which is based on logic analysis is presented toconstruct gene regulatory network. The method is validated on the gene expressionprofiles of263Saccharomyces cerevisiae deletion mutants. The power of this methodis demonstrated by comparing the deduced logic network with gene network reportedin the previous studies. Furthermore, interpretation of the deduced logic network leadsto the prediction of105targets and regulators of10transcription factors.