| Functional linkages describe pairwise relationships between proteins that work together to carry out a larger biological function. The function in which one protein is involved, when known, suggests the function of any unannotated partners. The challenge of combining these pairwise relationships into a unified description of protein function requires balancing the contribution of the linked proteins with each other and other sources of evidence. We have developed a Bayesian framework, known as Generalized Functional Linkages, to combine the information in functional linkages with existing knowledge about a protein, such as negative experimental results, its cellular location, or its individual molecular function. We demonstrate methods for defining functional linkages from protein-protein interaction networks in S. cerevisiae with an algorithm known as zorch, and with a phylogenetic profile method that addresses phylogenetic structure and operon predictions in prokaryotes. We also discuss a method for identifying proteins organized by mobile elements that confer function to prokaryotes through horizontal gene transfer, and an entropy-based Gene Ontology similarity metric that addresses both graph structure and the frequency of annotations. |
