Finding biologically accurate clusterings in hierarchical tree decompositions using the variation of information

Hierarchical clustering is a popular method for grouping together similar elements based on a distance measure between them. In many cases, annotations for some elements are known beforehand, which can aid the clustering process. We present a novel approach for decomposing a hierarchical clustering into the clusters that optimally match a set of known annotations, as measured by the variation of information metric. Our approach is general and does not require the user to enter the number of clusters desired. We apply it to two biological domains: finding protein complexes within protein interaction networks and identifying species within metagenomic DNA samples. For these two applications, we test the quality of our clusters by using them to predict complex and species membership, respectively. We find that our approach generally outperforms the commonly used heuristic methods.