Abstract
As gene annotation databases continue to evolve and improve, it has become feasible to incorporate the functional and pathway information about genes, available in these databases into the analysis of gene expression data, for a better understanding of the underlying mechanisms. A few methods have been proposed in the literature to formally convert individual gene results into gene function results. In this paper, we will compare the various methods, propose and examine some new ones, and offer a structured approach to incorporating gene function or pathway information into the analysis of expression data. We study the performance of the various methods and also compare them on real data, using a case study from the toxicogenomics area. Our results show that the approaches based on gene function scores yield a different, and functionally more interpretable, array of genes than methods that rely solely on individual gene scores. They also suggest that functional class scoring methods appear to perform better and more consistently than overrepresentation analysis and distributional score methods.