Extended Data Fig. 6: Molecular signatures and cell types define the characteristics of the subtypes.
a, How mRNA and protein signatures that defined Sub6 are selected and used for GSEA are shown as an example (n = 150 tumors). Integrative clustering (top left) shows that Sub6 was defined by RNA3, Prot5, and Phos5 clusters. Based on this information, 702 genes (rna3) defining RNA3 were selected as mRNA signatures (S6-G), and 1022 proteins (prot5) and 657 phosphopeptides (phos5) defining Prot5 and Phos5, respectively, were selected as protein signatures (S6-P) (bottom and top right). After mapping the phosphopeptides into phorphorylated proteins and combining them with 1022 proteins (prot5), the resulting proteins were used for GSEA. b, UMAP plot showing 9 clusters of the cell types indicated. Color coding was used to distinguish clusters. c, Genes specifically up-regulated in the 9 cell clusters. The heat map shows increased (yellow) and decreased (purple) expression (Z-score) of the genes up-regulated (row) in at least one cluster with respect to the median expression across all cells (column). Z-score was obtained from an inverse normal distribution of mRNA expression level. The color bar represents the gradient of the normalized Z-score. d, Violin plots showing distributions of mRNA expression levels for the selected representative genes across the 9 cell clusters. e, Distributions of tumor cellularity measurements determined by the methods indicated for TS1–4 and IS1–2 (n = 37, 18, 14, 33, 26, and 22 tumors for TS1-4 and IS1-2, respectively, in qpure cellularity, KRAS VAF, and ESTIMATE cellularity; n = 33, 15, 13, 30, 24, and 19 tumors for TS1-4 and IS1-2, respectively, in Histological cellularity). In violin plots, the line indicates the median value.
