Abstract
BACKGROUND: Genome-wide association studies have uncovered common variants at many loci influencing human complex traits, such as high-density lipoprotein cholesterol (HDL-C). However, the contribution of the identified genes is difficult to ascertain from current efforts interrogating common variants with small effects. Thus, there is a pressing need for scalable, cost-effective strategies for uncovering causal variants, many of which may be rare and noncoding. METHODS: Here, we used a molecular inversion probe target capture approach to resequence both coding and regulatory regions at 7 HDL-C-associated loci in 797 individuals with extremely high HDL-C versus 735 low-to-normal HDL-C controls. Our targets included protein-coding regions of GALNT2, APOA5, APOC3, SCARB1, CCDC92, ZNF664, CETP, and LIPG (>9 kb) and proximate noncoding regulatory features (>42 kb). RESULTS: Exome-wide genotyping in 1114 of the 1532 participants yielded a >90% genotyping concordance rate with molecular inversion probe-identified variants in ≈90% of participants. This approach rediscovered nearly all established genome-wide association studies associations in GALNT2, CETP, and LIPG loci with significant and concordant associations with HDL-C from our phenotypic extremes design at 0.1% of the sample size of lipid genome-wide association studies. In addition, we identified a novel, rare, CETP noncoding variant enriched in the extreme high HDL-C group (P<0.01, score test). CONCLUSIONS: Our targeted resequencing of individuals at the HDL-C phenotypic extremes offers a novel, efficient, and cost-effective approach for identifying rare coding and noncoding variation differences in extreme phenotypes and supports the rationale for applying this methodology to uncover rare variation-particularly noncoding variation-underlying myriad complex traits.