Hierarchical clustering of 1,593 samples based on genes belonging to Clusters D and G.

<p>Clustering was performed by using Spearman (columns) and Euclidean (rows) distance and average linkage metrics for both genes and samples. Genes from Clusters D and G are marked by blue and red respectively. Each sample belongs to one of the 8 molecular subtypes and is marked by a bar, colored based on the subtype: basal (red), claudin-low (black), ERBB2+ (yellow), luminal A (blue), luminal B (cyan), normal-like (green), molecular Apocrine (mApo; magenta), luminal-like (dark gray).