Model Fitting Tools for BioSPICE and the Systems Biology Workbench

As a contribution to the BioSPICE project, we have developed an optimization module, for the task of fitting kinetic rate constants to time series concentration data. The algorithms use, both local searches, such as the Levenberg-Marquardt, and simplex, as well as global search methods such as simulated annealing and real coded genetic algorithms. We provide a description of the modules in terms of the algorithms used, and the software components. We describe a few test cases which serve the purpose of demonstrating the software. We comment on determination of confidence limits, and issues related to observability of the fitted parameters.