Thursday, August 9, 2007 - 3:40 PM

COS 130-7: Genomic network modeling: An approach to predict net ecosystem processes from microbial community structure data

Diana Nemergut, David M Bortz, and Sasha C Reed. University of Colorado

Nitrogen fixation is an entirely microbial process with important ecosystem-level implications. Many abiotic factors are known to directly affect nitrogen fixation rates, including moisture, temperature, oxygen and nutrient availability. In addition, biotic controls are also important, and much recent work has focused on microbial community analyses of the organisms and/or genes involved in nitrogen fixation. However, in the natural world microbes rarely live and function in isolation; instead they typically form intimate associations with other life, allowing intricate biochemical conversions between organisms. Through these associations, the waste products of one microbe are often quickly consumed by another, and the net functioning of complex communities can be challenging to predict based merely on the presence or absence of a single organism or gene. Here, we present a microbial genomic network model to predict net rates of nitrogen fixation in complex microbial communities. We have created a matrix of possible metabolic reactions in organisms created using data from whole genome sequences and from the primary literature. By selecting different starting molecules (i.e., carbon substrates) we can compute the reaction rates for a steady-state equilibrium by solving a linear programming problem. This model allows the differentiation of metabolic conversions within and between cells, and can be used to both estimate ecosystem processes based on microbial community information, and to optimize microbial community composition for a processes of interest (e.g., hydrogen production, carbon sequestration, etc.). Thousands of environmental microbial communities have been characterized and the complete genome sequences of over 500 bacteria and archaea are now available in the public databases, thus the development of models to use these data to understand ecosystems processes is critical.