DNA sequencing technology is revolutionizing biology. DNA sequence data now needs to be translated, into functional information. The first step is the prediction of the function of individual gene products from their sequence through comparison with other known genes and genomes. The second step, and perhaps the more challenging one, involves the use of this functional information on individual genes to predict cellular functions resulting from their coordinated activity. In order to accomplish the objectives of this second step, model in silico organisms need to be built based on genomic data, and systems analysis methods developed to describe them so as to analyze, interpret, and predict the genotype-phenotype relationship. At present there is no computational infrastructure available to address these challenges. This project addresses this overall question by the following two methods: 1. Structural and steady state analysis methods will be developed to formulate organism-scale metabolic networks from annotated sequence data. These analysis methods will be presented within the Biology WorkBench (bioweb.ncsa.uiuc.edu). This capability will include the development of a graphical user interface that shows the metabolic map for a genomically defined metabolic genotype and the display of flux balance solutions thereon. Flux balance methods will be developed to study the characteristics and capabilities of defined genotypes, including the variation in the genotype, prediction of metabolic shifts, prediction of genome scale expression, and formulation of defined media, 2. To characterize the dynamic characteristics of multi-enzyme systems, methods will be developed to catalog, interpolate, and predict enzyme kinetic properties. These methods can be used to assign enzyme kinetic properties to annotated genome sequences. Methods will also be developed for temporal decomposition of a large set of simultaneous metabolic reactions. These methods will be able to determine the independent modes of motion in complex systems, give a physiological interpretation of these modes, and to predict the pool transformation matrix that can subsequently be used for model reduction for description on a selected time scale.
The results of this research will form the basis for the needed computational infrastructure for the formulation and testing of in silico metabolic representations of living cells. Basically, the result is the important metabolic genotype-phenotype relationship, and means to analyze, interpret, and predict it. These relationships will form the basis for metabolic engineering based on annotated genomic data.
