Impact of Rare Non-coding Variants on Human Diseases through Alternative Polyadenylation Outliers

Abstract

Abstract Although rare non-coding variants (RVs) play crucial roles in human complex traits and diseases, understanding their functional mechanisms and identifying those most closely associated with diseases continue to be major challenges. Here, we constructed the first comprehensive atlas of alternative polyadenylation (APA) outliers (aOutliers) from 15,201 samples across 49 human tissues. Strikingly, these aOutliers exhibit unique characteristics markedly distinct from those of outliers based on transcriptional abundance or splicing. This is evidenced by a pronounced enrichment of RVs specifically within aOutliers. Mechanistically, aOutlier RVs frequently alter poly(A) signals and splicing sites, and experimental perturbation of these RVs indeed triggers APA events. Furthermore, we developed a Bayesian-based APA RV prediction model, which successfully pinpointed a specific set of RVs with significantly large effect sizes on complex traits or diseases. A particularly intriguing discovery was the observed convergence effect on APA between rare and common cancer variants, exemplified by the combinatorial regulation of APA in the DDX18 gene. Together, this study introduces a novel APA-enhanced framework for individual genome annotation and underscores the importance of APA in uncovering previously unrecognized functional non-coding RVs linked to human complex traits and diseases.