DNA methylation and gene expression are important regulatory processes in cell proliferation, differentiation and many disease traits. Aberrant changes in methylation can alter gene expression and even be fatal in fetus development.
Gene expression regulation has been extensively used in deciphering numerous phenotype traits and uncovering potential targeted therapy for complex diseases. The association between observed molecular changes and phenotypes can be better detected if the analysis of gene expression can be combined with other genetic and epigenetic features.
We propose a new statistical framework to integrate gene expression and DNA methylation based on a variety of scoring schemes devised for evaluation of association strength. Our method utilizes a gene set enrichment approach to uncover relevant pathways.
The method was applied to two datasets, the breast cancer invasive carcinoma disease and the lymphatic/blood endothelial cells. Our statistical framework detected several previously identified disregulated pathways including Wnt and hedghog signaling pathways and DNA damage response in breast cancer. In addition, potential novel pathways have been suggested by our model.