Skip to main content
. Author manuscript; available in PMC: 2012 Jan 21.
Published in final edited form as: J Theor Biol. 2010 Oct 16;269(1):1–15. doi: 10.1016/j.jtbi.2010.10.001

Figure 6. Repertoire dendrograms and their confidence bounds under non-parametric measures of dissimilarity in CD4+ dataset.

Figure 6

Dendrograms under various dissimilarity measures obtained from the non-parametric analogue of the model (3.11) using agglomerative clustering and the complete link. The dendrograms in panels (A1), (B1), and (C1) were obtained using the point estimates of the dissimilarity matrix calculated from the data, whereas the most right ones (A3), (B3), and (C3) were obtained using upper 95% confidence bound on the Frobenius norm distribution of the dissimilarity matrix. The corresponding bootstrap estimates of the entire norm distribution are provided in the center panels (A2), (B2), and (C2) as density estimators, with 95% bound marked with a vertical line. Top (A) panels: hierarchical clusters based on the nonparametric version of the Morisita-Horn dissimilarity index (3.3). Center (B) panels: clusters based on the non-parametric version of the mutual information dissimilarity (3.6). Bottom (C) panels: clusters based on the values of the Shannon entropy function with no direct pairwise comparisons.