Comparative Analysis of Metabolic Pathways in Metagenomics

Metagenomics is the study of organismal DNA sequences obtained from an environmental sample. Metagenomic studies often compare samples taken from different locations (e.g., the human gut, lakes and oceans, soil) to assess significant differences in taxonomic makeup, genetic makeup, or function. Differences in function can be characterized by measuring the differential abundance of metabolic pathways found among the organisms in a sample. This is typically done by taking reads from the sample (short DNA sequences) and aligning them to genes in a database on the basis of sequence similarity. The genes code for proteins which are known to be catalytic enzymes involved in biochemical reactions in canoncial metabolic pathways. Previous studies have binned sequences into pathways by treating a pathway simply as a set of genes, ignoring its network topology. In this project, we explore techniques for comparing metagenomic samples that take both metabolic pathway structure and enzyme abundance into account. 1. Background Metagenomics (or environmental genomics) is the study of organismal DNA sequences obtained directly from an environmental sample. This relatively new field of genetic research enables studies of microorganisms that are not easily cultured in a laboratory, as well as studies of microorganisms in their natural environment. Metagenomics is benefitting from the emergence of next generation sequencing technologies (e.g., pyrosequencing), which enable large amounts of DNA sequencing to be done quickly and cheaply. A metabolic pathway is a series of biochemical reactions that occur inside a cell. Edges in the pathway represent catalytic enzymes, which are proteins encoded by genes. Nodes in the metabolic pathway represent molecular substrates, such as dietary vitamins and minerals and other small molecules like H2O. These edges (enzymes) and nodes (metabolites) together form a directed cyclic graph (metabolic pathway). The KEGG Metabolism PATHWAY database [1] stores a global reference pathway representing an ensemble of pathways that have been identified in many organisms. 2. Problem Statement Microbes function in highly interdependent communities, and microbial community processes are intimately related to ecosystem functionality. The recent advent of environmental genomics provides a way to systematically study the relationship between global community function and environment at the molecular level. Comparative metagenomics approaches have revealed significant variation in sequence composition, genome size, evolutionary rates, and metabolic capabilities in samples from qualitatively dissimilar environments, thus providing evidence for evolutionary adaptations. Comparing biological processes associated with functional categories based on KEGG pathways has become a popular and important approach because it captures the functional characteristics of different environments in an intuitive and meaningful way. Several studies have used this approach, such as the functional profiling of nine biomes [2], the quantification of the metabolic pathway adaptations of different marine samples [3], and the functional categorybased clustering of gut microbiomes from obese and lean twins [4]. However, these studies consider each functional metabolic pathway simply as a set of genes, discarding the network structure and the distribution of genes in the pathway. Subsequent data analysis compares only the aggregate abundance of genes in each functional pathway. Here are three examples where an aggregate abundance comparison would produce similar functional profiles for different metagenomic samples, but a network analysis would have better differentiating ability: (1) two samples have different parallel pathways, but their total abundance is the same (Figure 1a and 1b); (2) the genes in one sample are evenly distributed, but the other sample has high abundance only for some genes (Figure 1c and 1d); (3) one sample covers the complete pathway, but the other does not (Figure 1e and 1f). Figure 1. Toy examples of metabolic pathways recovered from metagenomic sequences. Green edges with numbers denote the abundance of genes. Graphs were generated with Graphviz software [5]. In this project, we seek to find variation between metabolic pathways derived from metagenomic samples by taking into account both pathway structure and enzyme abundance.