Fig. 3

Visualizing Subtyping results. a, b represent the distributions of the identified clusters; we can see from (a) that the samples are divided into 12 clusters with labels 0–11. As shown in (b), we can see that the label 10 is almost normal; LumB is the majority in labels 0, 2, 5, 6 and 8; LumA is the majority in label 11; Her2 is mostly distributed in label 7; and Basal is mostly distributed in label 1