AIGNET: A System That Infers Large Scale Genetic Networks

Abstract

Recent advances of technology in bioinformatics have made gene expression comprehensive and several approaches have been proposed to infer the genetic networks, using such gene data [2, 3]. We previously proposed a system named AIGNET (Algorithms for Inference of Genetic Networks) in which either of two completely different network models work independently [4]. One model is a static Boolean network model based on a multi-level digraph approach which can treat a large number of expression data and the other is a dynamic network model such as S-system [1] which can infer the genetic network including a group of interdependent genes. We have demonstrated that AIGNET can infer some class of simple but large-scale genetic networks. However, the reliability and efficiency of these network models obviously depends on the structure of the data given to the system. In the previously proposed AIGNET, it was suggested that these two models should work in a supplementary manner to cover the disadvantage and limitation of individual models. Therefore, to further improve the reliability and efficiency of inference of genetic networks, we developed a new AIGNET in which a combination of these two network models is implemented and works in a cooperative manner. We show that this new AIGNET system becomes more powerful at inferring the large scale genetic networks.