Regensburg 2016 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

BP: Fachverband Biologische Physik

BP 7: Coupled Problems in Biological Systems (Focus Session)

BP 7.3: Vortrag

Montag, 7. März 2016, 12:00–12:15, H45

Uncertainty analysis for dynamic models in systems biology — •Daniel Kaschek — Physikalisches Institut, Universität Freiburg, Deutschland

Dynamic models have gained increasing importance for the way we understand complex behavior in cell biology. Although the structure of such a model may be highly conserved between different cell types, experiments show that the parameter values are not. Therefore, reliable and efficient methods to determine parameter values from cell-type specific, time-resolved data are crucial for precise predictions.

Here, we present a collection of methods to determine parameter values, parameter uncertainty and uncertainty of prediction in non-linear dynamic models. Lie-group theory is employed to detect symmetries in the model and to eliminate structurally non-identifiable parameters. The profile-likelihood is introduced as an indispensable tool to determine parameter confidence bounds and explore the non-linear relationships between parameters. Also model predictions can be associated to special likelihood profiles and their uncertainty can thereby be accurately quantified. Finally, the method of Lagrangian multipliers is presented as a way to exploit the local structure of the likelihood, guide us quickly along the profile paths and make the likelihood-based methods even more efficient.