Software for the Optimization of Kinetic Rate Coefficients though Biochemical Pathways with Metabolic Isotopomers
Tatonetti, N.P.; Crook, S.M.; and Vermaas W.F.J.
School of Life Sciences, Arizona State University
Data available on the concentrations of metabolic isotopomers in various biochemical pathways which can be labeled with isotopes have prompted a need for a computational method of isolating the carbon flux throughout these biochemical pathways. In this work, we present software for determining the kinetic rate constants in a model reaction equation. First, the program generates a set of differential equations based upon a reaction pathway file, which contains a simple syntax for tracking the potentially labeled atoms through a reaction network. Once the set of differential equation has been generated, the program uses a genetic algorithm to optimize the unknown parameters. Genetic algorithms rely on the concepts of natural selection and use the selective pressure of fitness to determine parameters by comparing model simulation results to experimental data. The fitness is used to determine how to allow different sets of parameters to “reproduce” to produce a new generation of parameter sets. When referring to sets of real numbers, “reproduction” means using a linear recombination of different sets to produce “offspring” parameter sets. The program is able to isolate statistically significant sets of parameters after an average of only fifty generations. The software can also be adapted for use as an optimization package for other types of modeling paradigms such as other types of differential equation models or stochastic systems.