Figure 7: Hierarchical clustering reveals distinct localization of CNVs along the proximal–distal axis.
From: DNA copy number changes define spatial patterns of heterogeneity in colorectal cancer
(a) CNVs derived from the 68 samples within the tumour were clustered using agglomerative hierarchical clustering. The objective function used for clustering was the root sum of squared distances between sub-cluster points; the values of each point were the predicted copy number of each gene represented by the panel. Samples clustered into two distinct clusters: cluster 1 and cluster 2. (b) To visualize intra-tumour heterogeneity in terms of CNV, a 3D morphological tumour model was generated from slides stained with H&E. The distribution of the two clusters, from a, was superimposed onto this 3D model; cluster 1 is depicted in green and cluster 2 in violet. The clusters distinctively localized along the proximal–distal axis. The four images show the 3D model from different orthogonal perspectives.
