Skip to main content
Advertisement

< Back to Article

The activity of human enhancers is modulated by the splicing of their associated lncRNAs

Fig 3

Impact of elncRNA splicing on cis-gene regulation in the human population.

(A) LINC00886 (ENSG00000240875) is a multi-exonic elncRNA whose splicing is associated with genotype of the SNP variant (rs187202716), which is also associated to the expression level of an elncRNA target, LEKR1 (ENSG00000197980). (Top panel) Distribution of the Percentage-Spliced-In (PSI) of elncRNA splicing (LINC00886, red) and target expression (RPKM) (LEKR1, green) in samples across the population that carry different alleles of the SNP variant (rs187202716). Spearman’s rho and p-values are shown. (Bottom panel) Genome browser illustrating the genomic positions of the elncRNA (LINC00886, red), its target (LEKR1, green), and their associated SNP (rs187202716, black). (B) Four models of causal inference testing that predict the relationship between joint seQTL variant (black box) associations with the splicing (sQTL) of multi-exonic elncRNAs (red boxes) and the expression level (eQTL) of their target protein-coding genes (green boxes). Schematic representation of the models of joint seQTL associations: (1) the variants are independently associated with elncRNA splicing and target expression (independent model); (2) direct association between the variant and elncRNA splicing mediates the indirect association between that and target expression (causal model); (3) direct association between the variant with target expression mediates the indirect association between that and elncRNA splicing (reactive model); and (4) the causative interaction between elncRNA splicing and target expression is more complex (undecided model). Direct associations are depicted as solid lines and indirect associations as dash lines. (C) Scatterplot depicting causal inference testing local FDR associated with each the four models (as illustrated in B). Number and proportion of joint seQTLs are provided in brackets for each model. Dotted red lines denote significance threshold at local FDR < 0.1. (D) Proportion of joint elncRNA seQTLs causally or non-casually predicted to mediate target gene expression associated with splicing junctions located at the 5´ end (red) or 3´ end (grey) ends of the transcripts. Differences between groups were tested using a two-tailed Fisher’s exact test. *** p < 0.001.

Fig 3

doi: https://doi.org/10.1371/journal.pcbi.1009722.g003