Fig. 1

Evaluation of different clustering methods. Shown are a comparison of three clustering methods using different measures: a Davies–Bouldin index. b Silhouette index. c Difference in contact counts within and between clusters. d Number of clusters with an enriched regulatory signal. e − logP value of ANOVA for different regulatory signals. Each clustering method was applied with five different distance/similarity criteria between genomic regions: Euclidean, Pearson’s correlation, Spearman’s correlation, contact count, and log2 of contact count. The numbers (in red) correspond to the rank of the methods using the given measure. Cont. cnts contact count