VDJtools: Unifying Post-analysis of T Cell Receptor Repertoires
Fig 3
Overlap and clustering of TCR repertoires.
A. Hierarchical clustering of healthy donor and multiple sclerosis (MS) patient samples using F pairwise similarity metric (the geometric mean of the total frequency of overlapping clonotypes in first and second sample in pair). B. Multi-dimensional scaling (MDS) plot. Samples were projected onto two-dimensional plane based on pairwise similarities (F metric). C. Permutation testing for closeness of samples coming from the same group based on MDS plot. The plot shows observed (dashed red lines) and permuted (histograms) average within-group sample distance. In contrast to control group, MS group displays highly dissimilar T-cell repertoires. N = 10,000 permutations of group labels were performed. D. Hierarchical clustering of samples based on the Euclidean distance between Variable segment frequency vectors. Note that the clustering provides a nice separation between sample groups (Control and MS, P = 0.013, Fisher’s exact test).