2016 Conference on Computational Modelling with COPASI
Manchester Institute of Biotechnology, 12th – 13th May, 2016
1 - Newcastle University, UK
Keywords: Parameter Estimation, Identifiability Analysis, Profile Likelihood, Python Module
Parameter estimation is a difficult problem in systems modelling owing to high dimensionality and largely unknown topology of the systems being modelled. Identifiability analysis is a necessary part of parameter estimation to assess whether parameters can uniquely be determined by the defined optimization problem. It is not however always simple and straight forward to perform an identifiability analysis. The current work extends the work of Schaber (Biosystems 2012), which provides a method to calculate the profile likelihood method of identifiability analysis using the popular GUI based simulation engine, COPASI. The current work provides a tool that automates this procedure. Moreover the current work extends this method to facilitate the calculation of multiple profile likelihoods from an arbitrary number of parameter sets, indexed by rank of best fit. Lastly, facilities are provided to plot the results of the profile likelihood calculations and to automate the calculation of the associated likelihood ratio based confidence intervals. The use of the software is demonstrated using published models. Overall this work provides a tool to facilitate the estimation of parameters in systems biology using COPASI.