Comprehensive analysis of co-occurring domain sets in yeast proteins

Background: Protein domains are fundamental evolutionary units of protein architecture, composing proteins in a modular manner. Combinations of two or more, possibly non- adjacent, domains are thought to play specific functional roles within proteins. Indeed, while the number of potential co- occurring domain sets ( CDSs) is very large, only a few of these occur in nature. Here we study the principles governing domain content of proteins, using yeast as a model species. Results: We design a novel representation of proteins and their constituent domains as a protein-domain network. An analysis of this network reveals 99 CDSs that occur in proteins more than expected by chance. The identified CDSs are shown to preferentially include ancient domains that are conserved from bacteria or archaea. Moreover, the protein sets spanned by these combinations were found to be highly functionally coherent, significantly match known protein complexes, and enriched with protein- protein interactions. These observations serve to validate the biological significance of the identified CDSs. Conclusion: Our work provides a comprehensive list of co- occurring domain sets in yeast, and sheds light on their function and evolution.