New Concepts for Evaluating the Performance of Computational Methods

Research in Systems Biology is currently entering a new era. After a decade characterized by adopting existing experimental protocols and theoretical approaches to the requirements of Systems Biology, there is now a variety of tools and approaches available. However, many statistical and modeling concepts are not well-tested in application settings and their applicability is often seriously delimited. Therefore, a major challenge for the transfer of theoretical approaches to applications is the assessment and optimization of the methods performance for supporting experimental research. In this paper, new concepts for assessing methods winch were developed for anaIyzing experimental data in the context of systems biology will be introduced. Son Le ideas are illustrated by evaluating the impact of the logarithmic transformation for parameter estimation. A strong benefit of the log-transformation was observed for five different ODE models. The suggested framework enables less biased and more reliable and valid assessment and comparison of competing approaches than currently performed in the literature. The presented concepts could serve as basis for developing decision guidelines for optimal selection of analysis methods and thereby enhancing the transfer of systems biological procedures and reverse engineering methods to industrial applications like drug development. (C) 2016 IFAC (International Federation Of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.