We propose a methodology for the identification of transcription factors involved in the deregulation of genes in tumoral cells. This strategy is based on the inference of a reference gene regulatory network that connects transcription factors to their downstream targets using gene expression data. The behavior of genes in tumor samples is then carefully compared to this network of reference to detect deregulated target genes. A linear model is finally used to measure the ability of each transcription factor to explain those deregulations. We assess the performance of our method by numerical experiments on a breast cancer data set. We show that the information about deregulation is complementary to the expression data as the combination of the two improves the supervised classification performance of samples into cancer subtypes.