Pathway Analysis and Its Applications

Introduction

Recent technological developments in high-throughput genetic, genomic, and metabolomics profiling techniques has greatly expanded the potential for more systems level analysis. As such data becomes more readily available experimentally, and the scale of the data increases, this creates exciting new challenges in analyzing this “big data”. Handling and summarizing such high dimensional data in an efficient and interpretable way is crucial to making good use of this data. While there are a number of different strategies for handling big “-omics” data, pathway and network analysis approaches are becoming standard approaches for discovering and summarizing the underlying relationships in the data. The pathway and network approaches rely on either external knowledge bases or strong correlation structure in the data to collapse the data from thousands or millions of variables, to hundreds or thousands of pathways/networks for analysis. The results of such analyses are valuable to the process of discovering the underlying mechanism of disease or a phenotype of interest, including the events leading up to initiation, progression, and treatment of a disease.

As stated by the National Human Genome Research Institute (NHGRI), “A biological pathway is a series of actions among molecules in a cell that leads to a certain product or a change in a cell” (www.genome.org). Such a pathway can describe the function of molecules in a cell (regulatory pathway) or the change in chemical elements throughout the cell (metabolomics pathway) or a description of the initiation of a disease (disease pathway). It is now understood that biological pathways in the cell interact with one another to carry on the actions of the cell. Therefore, a group of interacting pathways comprise a biological network. There is a wealth of knowledge on such interactions that have been curated in pathway/knowledge bases that can be leveraged in statistical analysis of specific datasets. Many pathway based analysis approaches have been developed to use these databases to aid in gene function prediction, discover new associations with the trait of interest, and even to better classify patients or sample. Other approaches, more typically referred to as network approaches, focus on quantifying the connections between the gene, proteins, or metabolites to better understand the connections between the molecules that result in the phenotype/disease. A biological pathway and network can be displayed and analyzed in a graph form with vertices and edges. In such a form a vertex represents each element contained in the pathway or network and an edge represents an interaction (activation, repression, methylation, series of chemical reactions, etc.) between those elements. Figure 1 shows a graph representation of the Glycolysis Pathway, which displays many of the properties (scale-free degree distribution, high clustering coefficient, and characteristic path length) displayed in a variety of biological pathways. In the current chapter, we review many of the major categories of pathway and network analysis tools.

Figure 1.

Graph Representation of Glycolysis Pathway. Each vertex in the graph corresponds to an element (gene, SNP, metabolic process, etc.) and each edge corresponds to interactions between those elements. This representation was created in the Cytoscape software utilizing the KEGGscape plug-in. (http://www.cytoscape.org/)