Abstract 11946: Prioritization of Multiomic Rare Variants (RVs) Underlying Electrocardiogram (EKG) Traits Using Bayesian Hierarchical Modeling (“Watershed”)

Abstract

Introduction: Atrial fibrillation (AF) is the most common cardiac arrhythmia, affecting >33 million people worldwide. EKG is a powerful tool to diagnose AF. Earlier studies found common and rare genetic variants associated with EKG traits and AF risk. However, each individual genome has thousands of RVs which have not been systematically characterized in EKG traits. Hypothesis Individuals with outlier levels of omic measurements in EKG relevant genes are enriched with functional RVs which contribute to AF risk. Methods We analyzed transcriptomic, methylomic, proteomic, and whole genome sequencing data from 1319 individuals in TOPMed Multi-Ethnic Study of Atherosclerosis. We prioritized RVs with Watershed integrating genomic annotations and omic outlier signals (Fig A). We applied Watershed posterior probabilities as weights for EKG trait association using 30 million RVs in 4559 individuals. We developed a polygenic risk score (PRS) using RV posteriors to assess risk stratification for PR interval. Results The multiomic Watershed model prioritized RVs with large effects as identified by PR interval GWAS ( Fig B ). Each signal contributed to distinct sets of variants. Watershed confirmed known EKG gene associations and implicated splicing outliers in NDUFB8 as a novel association with QT interval. When incorporated in SKAT test, Watershed leveraged 10x more noncoding RVs and replicated associations between PR interval and SCN5A , MYH6 , NEBL , and GORASP1 (missed by the default SKAT model). An RV score synthesized across 494 genes was correlated with PR interval after adjusting for common variant PRS ( Fig C ), suggesting that Watershed-prioritized RVs can complement common variants to improve patient risk stratification. Conclusions Watershed is a promising framework to characterize RVs with functional effects, which informed trait-gene associations and improved patient risk stratification, enhancing mechanistic insights into EKG traits and AF diagnosis.