Extended Data Fig. 6: Correlations between reaction pattern rank, survival statistic, and gene dependency in lung adenocarcinomas and associated cell types.
From: Metaboverse enables automated discovery and visualization of diverse metabolic regulatory patterns
a. DepMap-generated plot of SMS and GLYCTK dependency scores selecting for lung adenocarcinoma cell lines only. b. Pearson correlation with loess regression of top 12 Average reaction patterns’ component enzymes and their genes’ survival statistics. The rank of each enzyme is plotted on the x-axis, and the corresponding survival BH statistic is plotted on the y-axis. The rank of an enzyme corresponds with the reaction pattern rank for which that enzyme is involved. Several of these reactions contained 2-5 gene components per enzyme (such as UPP1/2 complex (reactions #4 & 5) and alpha-amylase (reaction #9)). Reactions without a rank (reaction #8, sucrase-isomaltase dimer) did not have a corresponding survival statistic from the TCGA gene expression dataset to investigate. Shading indicates 95% confidence interval of regression. c. Dotplot of gene markers (horizontal axis) across annotated cell types (vertical axis) in the Human Lung Cell Atlas (HLCA) project. Proliferative gene markers were sourced from Whitfield, M., George, L., Grant, G. et al. Common markers of proliferation. Nat Rev Cancer 6, 99-106 (2006). The other genes displayed are based on the top Metaboverse hits’ corresponding enzymes from the LUAD metabolomics dataset. Dot size is scaled by fraction of cells in a cell type group expressing the given gene marker. Dot shade is scaled by mean expression of gene marker within the cell type group. Source numerical data are available64.
