Obtaining reliable predictions from large-scale dynamic models is a challenging task due to frequent lack of identifiability. This work presents a methodology for obtaining high-confidence predictions in biotechnological applications using metabolite time-series data. To preserve the complex behaviour of the network while reducing the number of estimated parameters, model parameters are combined in sets of meta-parameters, obtained from correlations between metabolite concentrations and between biochemical reaction rates. Next, an ensemble of models with different parameterizations is constructed and calibrated. Convergence of model outputs (consensus) is used as an indicator of confidence. Computational tests were carried out on a metabolic model of Chinese Hamster Ovary (CHO) cells. Using noisy simulated data, averaged ensemble predictions with high consensus were found to be more accurate than either predictions of individual ensemble models or averaged ensemble predictions with large variance. The procedure provides quantitative estimates of the confidence in model predictions and enables the analysis of sufficiently complex networks as required for practical applications in biotechnology.
CITATION STYLE
Villaverde, A. F., Bongard, S., Mauch, K., Müller, D., Balsa-Canto, E., Schmid, J., & Banga, J. R. (2014). High-confidence predictions in systems biology dynamic models. In Advances in Intelligent Systems and Computing (Vol. 294, pp. 161–171). Springer Verlag. https://doi.org/10.1007/978-3-319-07581-5_20
Mendeley helps you to discover research relevant for your work.