Nutrient-related Analysis of Pathway/Genome Databases

Abstact

We present an algorithm that solves two related problems in the analysis of metabolic networks stored within a pathway/genome database. (1) The Forward Propagation Problem: given a set of nutrients that are inputs to the metabolic network, what compounds will be produced by the metabolic network? (2) The Backtracking Problem: given the results of a forward propagation, and given a set of essential compounds that are not produced as a result of the forward propagation, what precursors must be supplied to produce those essential compounds? A program based on this algorithm is applied to the EcoCyc database, which is a pathway/genome database for E. coli that consists of annotated genomes and the metabolic reactions and pathways associated with the known gene products. The inputs to the program are a description of the metabolic network of an organism (EcoCyc), a set of nutrients corresponding to a known minimal growth medium, and a list of essential compounds to be produced. The program “fires” the microorganism’s metabolism contained in the database and predicts all synthesized and nonsynthesized essential compounds, along with the missing precursors required to produce the latter. When applied to the EcoCyc database, the program identifies a number of missing precursors that indicate incomplete regions of the database. Thus the program results can be used to evaluate existing pathway databases like EcoCyc.

Keywords: Artificial Intelligence, Artificial Intelligence Center, AIC