To generate functional gene groups for the analysis described in Fig. 2a and the text, we annotated all probes with data from Kyoto Encyclopedia of Genes and Genomes Pathways (http://www.genome.jp/kegg), the Gene Ontology project (http://www.geneontology.org), the Pfam database (http://www.sanger.ac.uk/Software/Pfam), mouse knockout phenotypes and human disease phenotypes collected by Kevin Becker’sgroup at the National Institute on Ageing24,25, the GSA project at Stanford (http://www-stat.stanford.edu/~tibs/GSA) and the GSEA project at the Broad Institute (http://www.broad.mit.edu/gsea), the HPRD project (http://www.hprd.org), as well as many groups collected from diverse sources at NCBI (http://www.ncbi.nlm.nih.gov), including protein-protein interactions and miRNA binding motifs26,27. Compilation of functional information from all of these sources and considering only gene groups of size 3–1,000 resulted in 23,810 partially redundant and overlapping functionally related gene groups. Enrichment of functional gene groups within various gene lists as described in the text was assessed by a standard hypergeometric test.
