Figure 3

(A) Principle Component Analysis of the RNA data showing clustering. panCK status outlined in green. (B) Unsupervised hierarchical clustering of the 1000 most variable genes across all datasets clustered together based on the IHC status. Green box shows distinct panCK differences from two samples in case 5. The blue box highlights differences in the S100β+ sample from case 5, while the red box shows the difference panCK-ve/S100β− samples. (C) Displays the difference in heterogeneity across topograpahy within the same case (case 5), across perivascular extension, perineural invasion, and tumour margin. (D) Heat map showing the range of heterogeneity within the same core, where two samples were taken across the same topographical areas.