Skip to main content
. 2022 Sep 21;11:e75227. doi: 10.7554/eLife.75227

Figure 3. Negative association between RNA- and protein-level regulation across genes and pathways.

(A) Dot plot showing the association between DNA–RNA and RNA–protein correlation, where each point is a gene. Density distribution is shown and a density distribution-dependent slope was calculated to estimate the association between the DNA–RNA and RNA–protein correlations. (B) A pathway-level analysis for the association between DNA–RNA and RNA–protein correlations (pan-cancer analysis, Clinical Proteomic Tumor Analysis Consortium [CPTAC]). The DNA–RNA (x-axis) and RNA–protein (y-axis) correlation for each cellular pathway was calculated using the median rho value across all genes belonging to the pathway (pathway database: msigdbr, v7.4.1, category = C5). Spearman’s (rho) and Pearson’s (r) correlation coefficients are shown at the top right of the panel. (C) Top panel: a heatmap showing Spearman’s correlations among different proteogenomic datasets. For each dataset, we first calculated the DNA–RNA and RNA–protein rho values for each gene, and then we calculated the Spearman’s correlation between these rho values (DNA–RNA rho or RNA–protein rho) across datasets. Bottom panel: a heatmap showing the pathway-level DNA–RNA and RNA–protein correlation score among different datasets. The pathway-level score was calculated by the median value across all genes in the same pathway and then Z-score transformed (pathway database: msigdbr, v7.4.1, category = C5). (D) A pathway-level analysis for the DNA–RNA (CPTAC) and RNA–protein (normal tissues, Wang et al., 2019) correlations. The DNA–RNA (x-axis) and RNA–protein (y-axis) correlation for each cellular pathway was calculated using the median rho value across all genes belonging to the pathway (pathway database: msigdbr, v7.4.1, category = C5). Spearman’s (rho) and Pearson’s (r) correlation coefficients are shown at the top right of the panel. (E) Density distribution of DNA–RNA correlations for genes belonging plasma membrane and RNA–protein correlations for genes belonging to ribosome, proteasome, mitochondria, and plasma membrane. The dashed line represents the specific cell compartment as indicated, the transparent purple or golden yellow line represents the median of complex (all cell compartments) or non-complex genes (all cell compartments). Significance between the genes in the specific cell compartment and complex or non-complex genes of all cell compartments was evaluated based on bootstrapping test and adjusted for false discovery rate (FDR). For example, the FDR in the top left panel was evaluated based on the difference between plasma membrane (non-complex) and complex genes from all cell compartments. (F) Gene-wise variability levels of scRNAseq data from Korean colorectal cancer patients (Lee et al., 2020) estimated by VarID (Grün, 2020). Genes were grouped according to their preferential regulation at the protein level (Group 1) or RNA level (Group 2). The averages of corrected variance estimates per gene are shown (‘Methods’). Box sizes represent the interquartile range (IQR), whiskers expand to± 1.5*IQR of the box limits, and outliers beyond the whisker limits are also shown. p-adjusted value: *<0.001 (one-sided Wilcoxon test with Bonferroni correction).

Figure 3.

Figure 3—figure supplement 1. Enriched gene sets among genes with different regulation in primary tumors and cell lines.

Figure 3—figure supplement 1.

(A) Top 10 Gene Ontology enriched gene sets of genes in Group 1 (high DNA–RNA correlation, low RNA–protein correlation; high protein-level regulation) in dark green and in Group 2 (low DNA–RNA correlation, high RNA–protein correlation; high RNA-level regulation) in light green among different cancer types (Clinical Proteomic Tumor Analysis Consortium [CPTAC]) (see text and ‘Methods’). Dot size represents the significance of the false discovery rate (FDR) as indicated. (B) Top 10 Gene Ontology enriched gene sets in Group 1 and 2 genes as (A) among different cancer cell lines (Cancer Cell Line Encyclopedia [CCLE]). Dot size represents the significance of the FDR. (C) A pathway-level analysis calculated and shown as Figure 3B but also displaying the standard error (SE) calculated among the genes in the same pathway. The pathway dot size and color represent the half-life of protein (h) and mRNA (h), respectively. The error bars represent standard error of the mean.
Figure 3—figure supplement 2. DNA–RNA and RNA–protein correlations among genes localized in different cellular compartments.

Figure 3—figure supplement 2.

Density plots of DNA–RNA and RNA–protein correlations for all cell compartments in Clinical Proteomic Tumor Analysis Consortium (CPTAC) pan-cancer analysis. Numbers at the right side of the figure represent the median of the correlation value of the density.