Skip to main content
. 2020 Aug 21;10(23):10531–10547. doi: 10.7150/thno.40944

Figure 2.

Figure 2

Classification of triple-negative breast cancers (TNBCs) by alternative polyadenylation (APA) profiling reveals 4 stable subtypes. (A) Consensus heatmaps showing the robustness of sample classification using non-negative matrix factorization (NMF) clustering. (B) Silhouette plot displaying the composition (n = number of samples) and stability (average width) of clustering. (C) Cophenetic and dispersion metrics for NMF across 2 to 10 clusters with 50 runs suggest 4 stable subtypes. (D) Principal component analysis depicts fundamental differences in the short 3′UTR index (SUI) between the TNBC APA subtypes. (E) The heatmap displaying the top 20 differentially polyadenylated 3′UTRs with most different SUIs in each subtype. Abbreviations: APA, alternative polyadenylation; NMF, non-negative matrix factorization; TNBC, triple-negative breast cancer; SUI, short 3′UTR index.