Network inference - Robust algorithms for regulatory network inference based on expression measurements and biological a prioir information

From 01-01-2003 to 31-12-2006

Description

In this project we like to reconstruct the regulatory network, active in the studied biological system, based on the gene expression measurements. De most important step in network inference is to learn as much as possible dependencies between gene and gene products from the raw expression data. As method of choice we opt for Bayesian networks which allow us to model joint probability distributions over a large collection of variables. Practically the goal of the project is to combine implementations of the most advanced algorithms in a platform that is usable inference program. This program is coupled with in-house developed machine learning tools to model the prior knowledge. The focus is on (a) the development of necessary extensions of Bayesian networks and (b) the thorough evaluation of the robustness of the inferences process. Given the high cost of the molecular experiments based on our results, the methodology should be very robust and reliable. Therefore we first like to develop a testgenerator from the data available in public databases on regulatory networks. Second we implement alternative techniques for parts of the process and compare the different results.

Team

• Kathleen Marchal, Coordinator
• Bart De Moor, Promoter
• Thomas Dhollander, Team member
• Olivier Gevaert, Team member
• Karen Lemmens, Team member
• Tim Van den Bulcke, Team member (External)

Financing

Funding: FWO - Research Foundation - Flanders

Program/Grant Type: FWO Research Grant - FWO Research Grant