Fig. 6

k-means clustering results based on the data from all four analytes and graph-theoretic analysis. a A 2D principal component analysis (PCA) plot illustrates the clusters. b The cluster size, follow-up time, and cases of death are listed. c Correlations of clusters with the tumor histology revealed that patients with a tumor histology other than the ductal type clustered together. d–g Number of alterations observed within the four analytes, grouped according to clusters formed based on all analytes. h Networks for each of the four clusters illustrate nodes with a degree ≥ 3 and directed edges between patients and parameters. The sizes of the nodes are proportional to the value of the (undirected) betweenness centrality and the intensity of the shade of green is proportional to the number of incoming edges. i Parameter nodes shown within the four networks were sorted based on their betweenness centrality. The 12 parameter nodes found in only one of the four networks were marked in light green. j The prevalence of signals/variants in the 12 unique parameter nodes (in i) was tabulated based on their occurrence in the four clusters. The cluster with the highest prevalence of a given parameter was marked in light blue and matched with the unique parameter nodes in i. Signals/variants of parameters found exclusively in just one cluster were marked in dark blue