Home
 
   Home
 General Info
   Overview
   People
   Visitor Info
   Links
   Impressum
 Research
   Topics
   Publications
   Awards
 Education
   Courses
   Thesis Projects
   eLearning
   Bulletin Board
   Student
     Exchange
   More...
 News
   Seminars
   Events
   In the Press
   Jobs
 
printable view back

Einladung zum Vortrag im Kolloquium Technische Kybernetik

Comparing Modeling Approaches and Evolutionary Algorithms on the Parameter Optimization of Metabolic Networks

Prof. Andreas Zell
Wilhelm-Schickard-Institute for Computer Science
Center for Bioinformatics (ZBIT)
Eberhard-Karls-Universität Tübingen

    Zeit: Freitag,· 11. 5. 2007 · 14:00 Uhr
    Ort: IST-Seminarraum 3.241 · Pfaffenwaldring 9 · Campus Stuttgart-Vaihingen

Abstract

Parameter estimation for biochemical model systems has become an important problem in Systems Biology. Here we focus on a metabolic subnetwork of the valine and leucine biosynthesis in the species C. glutamicum. We derived six alternative ordinary differential equation models based on the formalisms of the Michaelis Menten (MM) kinetics, the so called convenience kinetics (CK) and the generalized mass action kinetics (GMAK). We introduced two alternative modeling approaches for feedback inhibition to the latter formalism and evaluated the applicability of six optimization procedures (multi-start hill climber(MS-HC), binary and real valued Genetic Algorithm (binGA, realGA), standard and covariance matrix adaption Evolution Strategy (stdES, cmaES) as well as Simulated Annealing (SA)) to the problem of parameter fitting. We benchmarked the impact of different mutation and crossover operators as well as the influence of the population size or the parent/children ratio. Interestingly, the simple binGA performed best on average over all models, but the cmaES found the best result in total based on the relative squared error. The GMAK with the lowest number of model parameters provided the best fit. The convenience kinetics with the highest number of parameters showed a significantly smaller standard deviation from their local optimum and lead to the most stable systems.

Biographical Information

Dr. Andreas Zell is professor of computer science at Eberhard-Karls University of Tuebingen. He obtained the university diploma in computer science from the University of Kaiserslautern in 1986, and a M.S. degree from Stanford University, USA (1987). From 1986-1989 he was research assistant at the University of Stuttgart and in 1989 obtained his Ph.D. in computer science. From 1990-1994 he was assistant professor at the Institute for Parallel and Distributed Systems (IPVR) and obtained the venia legendi in 1994. Since 1995 he is full professor and chair of computer science at the University of Tuebingen. In 2000-2002 he was Dean of the computer science dept., and he is the founding director of the Centre for Bioinformatics Tuebingen (ZBIT), one of 5 national bioinformatics centres funded by the German national science foundation (DFG).


Weitere Informationen:
Prof. F. Allgöwer · Institut für Systemtheorie und Regelungstechnik · 0711 685 67738 · allgower@ist.uni-stuttgart.de
Uni logo Universität Stuttgart