Unsupervised clustering of 447 cutaneous melanoma patients and identification of the better prognostic group. (A) Integrative clustering (“Integrated”) based on three omics datasets (methylation, mRNA-seq and miRNA-seq) identified four clusters. Clustering analyses of methylation and mRNA-seq data divided patients into four clusters (“Methylation”) and three clusters (“mRNA”), respectively. Kaplan-Meier OS curves are shown for the clusters identified by the “Methylation” (B), “mRNA” (C) and “Integrated” (D) classifications. Patients were pooled into the better prognostic group if they exhibited significantly (P < 0.01) better OS in at least two clusterings. A similar identification technique was used for the poorer prognostic group. The final clustering identified 301 patients with better prognoses, 55 patients with poorer prognoses and 91 patients with intermediate prognoses (A), and the associated Kaplan-Meier OS curve is shown in (E).