Disease network construction and analysis A case study of Alzheimer s disease

Abstract

Each cell in a biological system requires thousands of proteins to perform newlinespecific tasks at specific times and locations to function correctly. Gene newlinevariants can occasionally prevent one or more proteins from functioning newlinecorrectly by making a protein malfunction or not be produced at all by newlinealtering the gene s instructions which produces it. A variant can impair newlinenormal development or result in a disease when it changes a protein essential newlineto the body. This information can be represented as a interaction network. newlineConstruction and analysis of Gene-Gene networks, constructed using gene newlineexpression data, are popular ways to understand the underlying mechanisms newlineof complex diseases. The major challenge with gene expression data with newlinea large number of variables(genes) and a comparatively very small number newlineof samples is to extract disease-related information, as the gene expression newlinedata contains a vast amount of redundant data and noise. newlineThis thesis focuses on constructing a statistically and biologically meaningful newlineAlzheimer s disease gene networks from the gene expression data and newlinethen the problem-specific analysis of the constructed disease gene networks. newlineAs the first contribution to the thesis, we have introduced a novel framework newlineto construct Alzheimer s disease gene networks. The framework uses newlinet-test, correlation, Gene Ontology categories machine learning techniques newlineto construct the disease gene network and to detect the potential biomarker newlinegenes. In the second contribution, we have used the proposed framework newlineto construct the stage-wise Alzheimer s disease gene networks and carried newlineout the community analysis. We have proposed a new stable community newlinediscovery algorithm (Neighbour-based community discovery algorithm) for newlinecommunity analysis. In the third contribution, we have analyzed the stagewise newlineAlzheimer s disease gene network to identify the genes that may be newlineresponsible for or play an important role in the disease progression. We newlineiii newlinehave introduced a new centrality measure to rank the genes according