A critical step towards understanding the impact of genetic variation on disease is through identification of regulatory elements in the DNA sequence and understanding how they act on target genes. Thousands of whole-genome data sets on histone modification, transcription factor occupancy, chromatin accessibility, RNA expression and other functional data have been generated in hundreds of human and mouse cell types. A major challenge is to develop advanced computational methods to exploit this rich source of information for identifying functional elements and understanding their roles in gene regulation and disease. Most existing methods are developed for studying either one mark or one cell type at a time. Frequently these methods are adopted to study multiple cell types via overly simplistic strategies, which are inadequate to capture the sophisticated interrelationship and heterogeneity of epigenetic signals between cell types. The few methods for exploiting multi-cellular epigenomics also do not scale well for studying a large number of cell types. In this talk, I will introduce a novel and computationally scalable method called IDEAS (http://stat.psu.edu/~yuzhang/IDEAS/) that aims to jointly analyze epigenetic data in many cell types simultaneously, and detect differential regulatory modules across both the genome and cell types. I will use independent experimental data to demonstrate the superior power of IDEAS over existing state-of-the-art methods, and I will further discuss its extensions.