Large heterogeneous expression data comprising a variety of cellular conditions hold the promise of a global view of transcriptional regulation. While standard analysis methods have been successfully applied to smaller data sets, large-scale data pose specific challenges that have prompted the development of new and more sophisticated approaches. This paper focuses on one such approach (the Signature Algorithm) and discusses the central challenges in the analysis of large data sets, and how they might be overcome. Biological questions that have been addressed using the Signature Algorithm are highlighted and a summary of other important methods from the literature is provided.