Identification Of Phenotype Associated Genes Based On Molecular Biological Networks

In biology, disease is a special phenotype for human and other organisms. Thecause of disease is a complex process which is involving in the coordination of manygenes, but it is difcult to identify the coordination by the traditional molecular biologicalexperiments. Recently, the rapid accumulation of biological knowledge and multi-levelhigh-throughput/omics0data, revolutionary changed the research paradigm on theproblem. Instead of only focusing on single molecular, researchers are gradually extend-ing their research to systematically analyze genome-wide bio-molecular interactions, i.e.bio-molecular networks. In this context, bio-molecular network, a powerful tool to studycomplex problems, enables systematically integration of high throughput biological dataand plenty of biological knowledge. Recently, a large number of biological networkshave been constructed, including protein interaction network, transcriptional regulationnetwork, signal transduction network, metabolic network and so on.In this dissertation, we used the molecular network as the primary tool and com-bined the microarray data to analyze and predict the cause genes of pathogen for cropsand human disease by focusing on the nodes of molecular network and dynamic changesof the whole molecular network. There are three parts in my thesis:(1) A novel systems biology approach is presented to predict pathogenic genes of F.graminearum based on molecular interaction network and gene expression data.With a small number of known pathogenic genes as seed genes, a subnetwork thatconsists of potential pathogenic genes is identified from the protein-protein inter-action network (PPIN) of F. graminearum, where the genes in the subnetwork arefurther required to be diferentially expressed before and after the invasion of thepathogenic fungus. Therefore, the candidate genes in the subnetwork are expectedto be involved in the same biological processes as seed genes, which imply that theyare potential pathogenic genes.(2) A novel approach is presented to predict disease genes and identify dysfunctionalnetworks or modules, based on the analysis of diferential interactions between dis-ease and control samples, in contrast to the analysis of diferential gene or proteinexpressions widely adopted in existing methods. As an example, we applied ourmethod to the study of three-stage microarray data for gastric cancer. We identifiednetwork modules or module biomarkers that include a set of genes related to gastriccancer, implying the predictive power of our method. The results on holdout val-idation data sets show that our identified module can serve as an efective modulebiomarker for accurately detecting or diagnosing gastric cancer, thereby validatingthe efciency of our method.(3) A comprehensive database, namely eFG (electronic resource for Fusarium gramin-earum) is developed to the community for further understanding this destructiveX pathogen. In eFG database, diferent kinds of/omics0data are collected, suchas genome data, functional annotations, pathway information, and so on. In partic-ular, some valuable derived functional genomics data are also imported into eFG,including protein subcellular localizations, protein-protein interactions and orthol-ogous genes. At present, eFG is freely available at http://csb.shu.edu.cn/efg/with auser-friendly and interactive interface.