Skip to Main content Skip to Navigation

Validation de modèles qualitatifs de réseaux de régulation génique: une méthode basée sur des techniques de vérification formelle

Grégory Batt 1
1 HELIX - Computer science and genomics
Inria Grenoble - Rhône-Alpes, LBBE - Laboratoire de Biométrie et Biologie Evolutive - UMR 5558
Abstract : Genetic regulatory networks control the development and functioning of living organisms. Given that most of the genetic regulatory networks of biological interest are large and that their dynamics is complex, understanding of their functioning is a major biological problem. Numerous methods have been developed for the modeling and simulation of these systems. Surprisingly, the problem of model validation has received until recently little attention. However, the validation step is all the more important that, in the context of the modeling of genetic regulatory networks, the modeled systems are complex and still imperfectly known.

In this thesis, we propose an approach for testing the validity of models of genetic regulatory networks by comparing the predictions obtained from the model with experimental data. More specifically in this work, we focus on a class of qualitative models of genetic regulatory networks defined in terms of piecewise-linear (PL) differential equations. While having a simple mathematical form that favors their symbolic analysis, these models capture essential aspects of genetic regulations. Also, we would like to use the qualitative information on the dynamics of the system that consists in the time evolution of the direction of change of the concentrations of the proteins in the network. This information can be experimentally obtained from temporal expression profiles.

The method that we propose must satisfy two constraints. Firstly, it should make it possible to obtain predictions that are well-adapted to comparison with the type of data we consider. Secondly, given the size and the complexity of the networks of biological interest, it should also make it possible to check efficiently the consistency between predictions and observations.

To meet these two constraints, we extend in two directions an approach previously developed by de Jong and colleagues for the symbolic analysis of qualitative PL models. Firstly, we propose to use a finer-grained representation of the state of the system, allowing us to obtain, using discrete abstraction, predictions that are better-adapted to the comparison with experimental data. Secondly, we propose to combine this method with model-checking techniques. We demonstrate that the combined use of discrete abstraction and model checking makes it possible to check efficiently the dynamical properties, expressed in temporal logic, of continuous models.

This method has been implemented in a new version of the tool Genetic Network Analyzer (GNA 6.0). GNA 6.0 has been used for the validation of two models, that are large and complex, of the initiation of the sporulation in B. subtilis and of the nutritional stress response in E. coli}. We have thus verified that the predictions obtained from these models are consistent with most of the experimental data available in the literature. Several inconsistencies have also been identified, suggesting either model revisions or the realization of complementary experiments. In addition to a contribution to a better understanding of the functioning of these systems, these two case studies illustrate more generally that using the method we propose it is possible to test whether the predictions obtained from complex models are consistent with a variety of experimentally-observed properties.
Document type :
Complete list of metadata

Cited literature [222 references]  Display  Hide  Download
Contributor : Gregory Batt <>
Submitted on : Sunday, March 26, 2006 - 6:21:38 AM
Last modification on : Tuesday, July 20, 2021 - 5:20:02 PM
Long-term archiving on: : Monday, September 17, 2012 - 1:00:43 PM


  • HAL Id : tel-00012040, version 1



Grégory Batt. Validation de modèles qualitatifs de réseaux de régulation génique: une méthode basée sur des techniques de vérification formelle. Modélisation et simulation. Université Joseph-Fourier - Grenoble I, 2006. Français. ⟨tel-00012040⟩



Record views


Files downloads