Genetic variants regulating RNA splicing and transcript usage have been implicated in both common and rare diseases. However, identifying specific transcriptional effects of these variants remains challenging, partly because reference transcriptomes are incomplete and contain many truncated transcripts that lack annotated 3ʹ or 5ʹ ends. We developed a novel analytical approach to overcome these limitations by stratifying transcript annotations into three separate components: promoters, internal exons and 3ʹ ends. We apply our method to genotype and RNA-seq data from human macrophages exposed to a range of inflammatory stimuli (IFNɣ, Salmonella, IFNɣ + Salmonella) and a metabolic stimulus (acetylated LDL), obtained from up to 84 individuals. We found that over half of the quantitative trait loci (QTLs) colocalising with complex traits were identified only at the transcript level with no detectable effect on total gene expression. Furthermore, 55% of the transcript-level associations regulated either promoter or 3ʹ end usage, many of which are missed by methods that only quantify exon-exon junctions. Finally, we demonstrate that promoter-usage QTLs have distinct genetic architecture and are 50% more likely to be context-specific than alternatively spliced internal exons. In summary, we highlight how different RNA-seq quantification approaches capture distinct aspects of transcription and that characterizing the full spectrum of transcriptional consequences of genetic variation requires a combination of analytical strategies.

Genetic effects on promoter usage are highly context-specific and contribute to complex traits