Team:Paris/Characterization

In the Analysis section, we developed a model based on available experimental data and extended previously published models. This model has proven useful for the initial design stage, in the sense that it has suggested ways to improve the oscillatory behavior.

However, it relies on parameters collected from various sources and obtained under different conditions. So, this model based on bibliography has probably limited predictive capabilities. The challenge we address here is to obtain a predictive model of our system. Such a model would be a unique tool to tune and optimize the behavior of our system.

This model is constructed using a bottom-up approach. Each part is experimentally characterized using specific constructs. Key parameters are then obtained by fitting data to simple models of each part. These models should capture key features of gene regulation and must only involve experimentally measurable parameters. We consequently used Hill-type-like ODE models. However, to establish these models, we started from molecular reactions to obtain our non-linear model so as to explicit underlying assumptions and possible limitations. The overall model is then obtained by using a compositionality assumption: the model of the full system is obtained using models of each parts.

It is important to note that we do not follow the Standard Promoter Unit approach in which each part is characterized in normallized conditions to improve reusability of the characteisations across different labs. In contrast, our characterisations are made in conditions that are as close as possible to the experimental conditions in which our system is supposed to work (same strain, same growth conditions, same plasmid...). We stress that the goal in our case is to improve predictability of the resulting model

This work essentially amounts to quantifying promoter activities as a function of its transcription factors. Because many promoters need to be characterized, we designed a workflow that allows to carry out constructs, experiments and parameter estimation in a rational approach.

.
 * The whole approach is illustrated with a simple example>