libRCGA: a C library for real-coded genetic algorithms for rapid parameter estimation of kinetic models

  • Maeda Kazuhiro
    Frontier Research Academy for Young Researchers, Kyushu Institute of Technology Department of Bioscience and Bioinformatics, Kyushu Institute of Technology
  • Boogerd Fred C.
    Department of Molecular Cell Biology, Faculty of Science, VU University Amsterdam
  • Kurata Hiroyuki
    Department of Bioscience and Bioinformatics, Kyushu Institute of Technology Biomedical Informatics R&D Center, Kyushu Institute of Technology

Abstract

<p>Kinetic modeling is a powerful tool to understand how a biochemical system behaves as a whole. To develop a realistic and predictive model, kinetic parameters need to be estimated so that a model fits experimental data. However, parameter estimation remains a major bottleneck in kinetic modeling. To accelerate parameter estimation, we developed a C library for real-coded genetic algorithms (libRCGA). In libRCGA, two real-coded genetic algorithms (RCGAs), viz. the Unimodal Normal Distribution Crossover with Minimal Generation Gap (UNDX/MGG) and the Real-coded Ensemble Crossover star with Just Generation Gap (REX star/JGG), are implemented in C language and paralleled by Message Passing Interface (MPI). We designed libRCGA to take advantage of high-performance computing environments and thus to significantly accelerate parameter estimation. Constrained optimization formulation is useful to construct a realistic kinetic model that satisfies several biological constraints. libRCGA employs stochastic ranking to efficiently solve constrained optimization problems. In the present paper, we demonstrate the performance of libRCGA through benchmark problems and in realistic parameter estimation problems. libRCGA is freely available for academic usage at http://kurata21.bio.kyutech.ac.jp/maeda/index.html.</p>

Journal

Citations (4)*help

See more

References(47)*help

See more

Related Projects

See more

Details

Report a problem

Back to top