Skip to main content
Fig. 6 | BMC Genomics

Fig. 6

From: LABRAT reveals association of alternative polyadenylation with transcript localization, RNA binding protein expression, transcription speed, and cancer survival

Fig. 6

Comprehensive analyses of connections between alternative polyadenylation and transcript expression. (A) Diagram of correlation between APA and transcript expression. Rho (ρ) is defined as the correlation between changes in gene expression and changes in ψ value across two conditions. In the scenario described in the top row, the overall RNA expression level for the gene is high in sample A but low in sample B while the gene’s ψ value is low in sample A and high in sample B. Changes in gene expression and Δψ are therefore negatively correlated, giving ρ a negative value. Conversely, in the scenario described in the bottom row, changes in gene expression and ψ are positively correlated. (B) P values across all expressed genes within a comparison for the ENCODE RBP knockdown data. Each dot represents a single comparison (RBP knockdown vs control knockdown). P values for the correlation between gene expression and APA are indicated by dot shape and color. (C) P values across all expressed genes with a comparison for the TCGA paired tumor/control sample data. Each dot represents a single patient’s tumor and control samples. P values for the correlation between gene expression and APA are indicated by dot shape and color. (D) Gene-level ρ values across all ENCODE RBP knockdown experiments. (E) Gene-level ρ values across all TCGA tumor/control sample pairs. (F) Correlation of gene-level ρ values derived from the ENCODE and TCGA datasets (D and E). Red lines indicate the density of points, and the locations of three genes selected for further study are indicated by labels. (G) Correlation between gene expression changes and Δψ for three genes. Orange dots represent ENCODE sample pairs (RBP knockdown vs. control knockdown) while purple dots represent TCGA sample pairs (tumor vs. control samples). (H) Top: illustration of the UTR fragments fused to the Firefly luciferase gene. Bottom: RT-qPCR-derived relative levels of firefly luciferase mRNA expression when the proximal and distal UTR fragments of the indicated genes were fused. Values indicate ratios between the abundances of Firefly and Renilla luciferase mRNAs with this ratio in the proximal UTR comparison set to 1. P values were calculated using a Wilcoxon ranksum test. (I) Correlation between gene expression changes and Δψ was used to define positively correlated, negatively correlated and control genes with two APA isoforms. Correlations are calculated for ENCODE and TCGA separately. (J) Distal UTR lengths of each gene set. P values were calculated using a Wilcoxon ranksum test. (K) Distal UTR GC content of each gene set. P values were calculated using a Wilcoxon ranksum test. (L) Five-mer enrichments in the distal 3′ UTRs of positively and negatively correlated gene sets vs control. Five-mers are significantly enriched (BH-adjusted p < 0.05, Fisher’s exact test) in either both comparisons, one comparison or neither and are represented by a circle plus, open circle or closed dot respectively. Five-mers are colored by their AU content as ranked 0–5. Canonical AU rich element (ARE) “AUUUA” is highlighted as enriched in negatively correlated distal UTRs. (M) RBP motif enrichments in the distal 3′ UTRs of positively and negatively correlated gene sets vs control. RBP motifs are significantly enriched (BH-adjusted p < 0.05, Fisher’s exact test) in either both comparisons, one comparison or neither and are represented by a green circle plus, blue open circle or purple dot respectively. Canonical ARE binding protein motifs are highlighted as enriched in negatively correlated distal UTRs. (N) Distal UTR AREScores of each gene set as calculated by AREScore software. P values were calculated using a Wilcoxon ranksum test

Back to article page