Inferring regulatory networks through orthologous gene mapping

In recent years, constructing regulatory networks using gene expression data has received extensive attentions. From Boolean network, Bayesian network to Module network, a number of models have been applied in order to learn the regulatory networks more accurately. The statistical power of network modeling is directly affected by sample size of available expression data used as training data. However, training data are not always abundantly available, except a few well-studied model organisms. It is also infeasible to perform a large number of experiments which require a lot of resources and labor. How to learn a reliable network using minimal training data making use of well-characterized model organisms becomes an important problem with pressing needs. In this paper, we developed a method that infers regulatory sub-networks for a species with limited expression data by learning from a known reference network through orthologous gene mapping. Inspection of three predicted sub-networks confirms biological relevance of our predictions and demonstrates the ability of the method in extracting core regulatory relationships.