Skip to main content
Fig. 4 | BMC Genomics

Fig. 4

From: svclassify: a method to establish benchmark structural variant calls

Fig. 4

Hierarchical clustering results using L1 distance and Ward’s method shown as (a) a dendrogram and (b-c) in multi-dimensional scaling plots. (a) The horizontal dotted red line shows the cut-off at a cluster dissimilarity index of about 10000, which results in 8 clusters. The clusters are number 1 to 8 from left to right, with 4 and 5 containing primarily non-SVs, 8 containing a mixture of SVs and non-SVs, and 1, 2, 3, 6, and 7 containing different types of deletions (see Table 4). (b-c) Multidimensional scaling plots for visualizing the 8 clusters. We use a 3 dimensional representation of the data space which associates 3 MDS coordinates to each site, one for each dimension. (b) Plot of MDS-2 against MDS-1, which clearly separates Cluster 6 (mainly SVs with inaccurate breakpoints). (c) Plot of MDS-3 against MDS-1, in which the different types of SVs are generally well-separated from each other and from non-SVs

Back to article page