Figure 1
From: Molecular subtypes of breast cancer are associated with characteristic DNA methylation patterns
Unsupervised clustering of 189 tumours based on the 332 most variably methylated CpGs. (a) Hierarchical clustering. The heatmap shows relative methylation levels (red, more methylated; green, less methylated). Clustering results in three clusters associated with lumB, lumA and basal-like tumours, respectively. (b) Kaplan-Meier demonstrating longest survival in lumA-associated Cluster 2 and shortest in basal-like-associated Cluster 3. P-value was calculated using log-rank test. (c) Fraction of genome altered (FGA) highest in basal-like-associated Cluster 3 and lowest in lumA-associated Cluster 2. P-value was calculated using analysis of variance (ANOVA). (d) S-phase fraction highest in basal-like-associated Cluster 3 and lowest in lumA-associated Cluster 2. P-value was calculated using ANOVA. The number of tumours in each subtype is shown at top.