Skip to Main Content
The appearance of microarray technologies makes it possible to simultaneously monitor the expression levels for tens of thousands of genes. Bayesian Network (BN) is an important approach for predicting gene regulatory networks from expression data. However, two fundamental problems greatly reduce the effectiveness of current BN methods. The first problem is much less samples than genes, it makes that to find a "best" network is very difficult. The second is the excessive computational time, the search space of possible Bayesian network is very large because genes are numerous. In this paper, we introduce a new method to learn Bayesian networks which combines fuzzy clustering algorithm to reduce the search space. From the view of systems biology, modularity and hierarchy are key features of biological networks. This allows us to gain global network by assembling local components. A local component is composed of genes with same function. Those genes may have same or very similar expression pattern. Besides, those inter-related genes may be bridge of different functional modules. Consequently, in a process of learning of local networks, the gap between the number of samples and genes is shortened, and the search space is reduced.Our approach is evaluated using artificial data and expression data measured during the yeast cell cycle. Results demonstrate that this approach can predict regulatory networks with significantly improved accuracy and reduced computational time compared with existing BN approaches.