Fig. 2

NES clustering analysis defines four phenotypes of primary breast cancer with different survival outcomes. A Unsupervised hierarchical cluster (Ward.D2 criterion) of primary TCGA breast cancer samples applied to the matrix of the enrichment scores calculated through the single-sample Mann–Whitney–Wilcoxon gene set test for each identified mammary tumor prognostic onco-signature. In red the onco-signatures significantly enriched, and in blue those significantly depleted across the four clusters. B Kaplan–Meier overall survival curves of TCGA primary breast cancer dataset according to the cluster assignment. C Kaplan–Meier overall survival curves of TCGA primary Luminal breast cancer cohort determined by comparing Cluster 1 vs Cluster 2 vs Cluster 4. D Kaplan–Meier overall survival curves of TCGA primary Luminal breast cancer dataset determined by comparing Cluster 2 vs Cluster 4. E Gene ontology enrichment analysis of the differentially expressed genes between Cluster 4 and Cluster 2