Fast, Robust Clustering Algorithms for Gene Enrichment Data

Clusters functionally related biological terms from gene set enrichment results. Terms are compared by the overlap of their gene sets using Cohen's kappa, the Jaccard index, or the Dice coefficient, and the resulting similarity matrix is grouped either by agglomerative hierarchical clustering with single, complete, average, or Ward linkage, or by the seed-and-merge procedure of the 'DAVID' functional classification tool. The distance and clustering routines are written in 'C++' for speed. The methods are described in Huang et al. (2007) , Ward (1963) , Cohen (1960) , and Jaccard (1912) .