To elucidate cellular machinery on a global scale, we performed a multiple comparison of the recently available protein-protein interaction networks of Caenorhabditis elegans, Drosophila melanogaster, and Saccharomyces cerevisiae. This comparison integrated protein interaction and sequence information to reveal 71 network regions that were conserved across all three species and many exclusive to the metazoans. We used this conservation, and found statistically significant support for 4,645 previously undescribed protein functions and 2,609 previously undescribed protein interactions. We tested 60 interaction predictions for yeast by two-hybrid analysis, confirming approximately half of these. Significantly, many of the predicted functions and interactions would not have been identified from sequence similarity alone, demonstrating that network comparisons provide essential biological information beyond what is gleaned from the genome.comparative analysis ͉ multiple alignment ͉ protein network ͉ yeast two-hybrid A major challenge of postgenomic biology is to understand the complex networks of interacting genes, proteins, and small molecules that give rise to biological form and function. Advances in whole-genome approaches are now enabling us to characterize these networks systematically, by using procedures such as the two-hybrid assay (1) and protein coimmunoprecipitation (2) to screen for protein-protein interactions. To date, these technologies have generated large interaction networks for bacteria (3), yeast (4-7), and, recently, fruit fly (8) and nematode worm (9).The large amount of protein interaction data now available presents opportunities and challenges in understanding evolution and function. Such challenges involve assigning functional roles to interactions (10), separating true protein-protein interactions from false positives (11), and, ultimately, organizing large-scale interaction data into models of cellular signaling and regulatory machinery. As is often the case in biology, an approach based on evolutionary cross-species comparisons provides a valuable framework for addressing these challenges. However, although methods for comparing DNA and protein sequences have been a mainstay of bioinformatics over the past 30 years, development of similar tools at other levels of biological information, including protein interactions (12-14), metabolic networks (15-17), or gene expression data (18)(19)(20), is just beginning.Recently, we devised a method called PATHBLAST (13) for comparing the protein interaction networks of two species. Just as BLAST performs rapid pairwise alignment of protein sequences (21), PATHBLAST is based on efficient alignment of two protein networks to identify conserved network regions. Here, we extend this approach to present a computational framework for alignment and comparison of more than two protein networks. We apply this multiple network alignment strategy to compare the recently available protein networks for worm, fly, and yeast, and show that although any single net...