Abstract
High-throughput technology enables the genome-wide study in almost every field of biology. However, the majority of current researches are focused on the “single gene” approach, neglecting the information about the correlation structures across genes. In this study, we take into account the correlation structure using modules as our study units. We first find the functional gene modules and then identify the corresponding SNPs by Sparse Partial Least Square method for each module. Finally, the Bayesian network is built on both the genes within the modules and the SNPs underlying the modules to re-construct the pathways associated with the well-known cancer genes in order to realize the causal relationships within the gene regulatory network.