Fig. 2
From: Widespread intronic polyadenylation diversifies immune cell transcriptomes
Enrichment of IpA sites at the start of transcription units. a The fraction of retained coding region (CDR) was calculated as the nucleotides from the start codon to the end of the exon located upstream of the IpA peak, divided by all coding nucleotides of the longest annotated open reading frame and is shown for all IpA isoforms in the atlas. b As in a, but shown for individual cell types. c Correlation between the median retained CDR with the fraction of genes that generate IpA isoforms in each sample (Pearson's correlation coefficient, r = −0.86). Tissues with a higher proportion of IpA isoforms generate IpA isoforms with shorter CDRs. d Correlation between the median retained CDR and the median usage of the distal ApA site in the 3ʹUTR (Pearson's correlation coefficient, r = 0.60). Tissues with shorter 3ʹUTRs have IpA isoforms with shorter CDRs. e IpA isoforms occur in long introns. The introns in which 5ʹIpA events occur are longer than the other introns of the same genes (one-sided Wilcoxon rank-sum test, P < 10−20). Similarly, the introns in which 3ʹIpA events occur are longer than the remaining introns of those genes (one-sided Wilcoxon rank-sum test, P < 10−20). If taken together, then the introns in which IpA events occur are longer than the introns of the genes that only express full-length isoform (one-sided Wilcoxon rank-sum test, P < 10−20). f IpA isoforms occur in genes with long transcription units. Genes that express IpA isoforms have longer transcription units compared to genes that only express full-length isoforms (one-sided Wilcoxon rank-sum test, P < 10−20). g Higher conservation around the cleavage sites of IpA isoforms. The plot shows PhastCons scores of 200 nt upstream and downstream of IpA cleavage sites (x = 0). 5ʹIpA and 3ʹIpA events both have significantly higher conservation flanking the cleavage site compared to corresponding regions of randomly selected polyA signals (AAUAAA) in introns lacking IpA events (one-sided Wilcoxon signed-rank test, P < 10−68 for both comparisons). h Genes with IpA isoforms (n = 3481) are enriched for polyadenylation sites compared with genes that do no generate IpA isoforms (n = 12,092) (one-sided Wilcoxon signed-rank test, P < 10−18). The frequency of polyadenylation sites was counted from the TSS to the beginning of the 3ʹUTR and is shown as the average number of signals per kb per gene. i As in h, but U1 binding sites are shown. IpA genes are depleted for U1 snRNP signals (one-sided Wilcoxon signed-rank test, P < 10−18)
