Figure 4

Distance between the two clusterings created by different methods with the same input. Only the complete linkage for the hierarchical clustering is considered. The solid curves represent the clustering based on original gene expression data while the dotted curves stand for clustering based on spectral densities. Abbreviations are exploited for the conciseness of labels as follows: hier (hierarchical clustering), euc (Euclidean), cor (correlation), psd (power spectral density), exp (expression data).