Meta Analysis of Gene Expression Data Using Gene Regulatory Pathways

Date of Completion

January 2011


Biology, Bioinformatics|Computer Science




Using current experimental techniques in systems biology, it is possible to capture considerable quantitative details pertaining to genes in an experimental context. Correspondingly, gene regulatory networks are a great abstraction of involvement of genes in a system of interactions within a cell. There is a compelling need to present information from these complex experiments in a visually intuitive and computationally sound manner in order to extract biologically significant inferences about gene interaction and pathway involvement. Our overall objective is to formulate and implement a software system that allows consistency evaluation of experimental data with published or user generated gene regulatory networks in order to elucidate the pathways that are significantly involved and to facilitate interpretation of principles under which gene interactions occur within the assay. The proposed framework supports spraying of gene expression data over gene networks and algorithms are implemented that allow computation of a defined consistency measure for the given set. A coupled p-value is generated for each experiment/pathway set using statistical methods that enable ranking of pathways that are most consistent. The effectiveness of this approach in isolating relevant consistent KEGG pathways with a significant p-value in application to a number of studies employing varying experimental platforms is demonstrated by examples. While evaluating the consistency, the system is also able to make predictions for the expression level of elements in the network with missing values. The accuracy of such predictions is evaluated and reported using synthetic as well as real expression and network data. The main contribution lies in the development of a comprehensive software framework that incorporates contextual information published in gene regulatory networks to evaluate consistency and to make gene expression predictions. ^