Model identifiability and taste modelling

In the first part of this project the focus is on general tools for parameter identifiability and model reduction of large state space models (sets of ordinary differential equations). We start from a recently developed algorithm to address the issue of identifiability. This new algorithm will be explored further and applied to several (general) Systems Biology case studies.

In the second phase, the project will focus on model development for taste panel scores and the identification of a reliable model that can predict these scores. Uncertainty propagation and sensitivity analysis will form an essential part of this study. In this part of the project the PhD will closely collaborate with an experimental group within Bioscience developing the so-called FlavorTaster, a device that mimicks the human taste system via an array of taste and olfactory receptors. The PhD has as task to develop a model that can be used for taste panel predictions on basis of FlavorTaster sensor readings. In this project advanced and recently developed optimization techniques will be applied, both for parameter estimation and optimal experimental design, that allow the development of a reliable flavour prediction model. The project is a collaborative effort of physicists, nutricionists, and applied mathematicians.


Figure. Typical response trace profiles for a receptor array expressing three different receptors. They were challenged with a peptide specific only for the NK1 receptor (left panels) or a bitter compound only recognized by the bitter receptor (right panels). Responses are quantitative and specific.