Abstract
Whole genome sequencing studies applied to large populations or biobanks with extensive phenotyping raise new analytic challenges. The need to consider many variants at a locus or group of genes simultaneously and the potential to study many correlated phenotypes with shared genetic architecture provide opportunities for discovery and inference that are not addressed by the traditional one variant-one phenotype association study. Here we introduce a model comparison approach we refer to as MRP for rare variant association studies that considers correlation, scale, and location of genetic effects across a group of genetic variants, phenotypes, and studies. We consider the use of summary statistic data to apply univariate and multivariate gene-based meta-analysis models for identifying rare variant associations with an emphasis on protective protein-truncating variants that can expedite drug discovery. Through simulation studies, we demonstrate that the proposed model comparison approach can improve ability to detect rare variant association signals. We also apply the model to two groups of phenotypes from the UK Biobank: 1) asthma diagnosis (43,626 cases), eosinophil counts, forced expiratory volume, and forced vital capacity; and 2) glaucoma diagnosis (5,863 cases), intra-ocular pressure, and corneal resistance factor. We are able to recover known associations such as the protective association between rs146597587 in IL33 and asthma (log10 (Bayes Factor) = 29.4). We also find evidence for novel protective associations between rare variants in ANGPTL7 and glaucoma (log10 (Bayes Factor) = 13.1). Overall, we show that the MRP model comparison approach is able to retain and improve upon useful features from widely-used meta-analysis approaches for rare variant association analyses and prioritize protective modifiers of disease risk.
Author summary Due to the continually decreasing cost of acquiring genetic data, we are now beginning to see large collections of individuals for which we have both genetic information and trait data such as disease status, physical measurements, biomarker levels, and more. These datasets offer new opportunities to find relationships between inherited genetic variation and disease. While it is known that there are relationships between different traits, typical genetic analyses only focus on analyzing one genetic variant and one phenotype at a time. Additionally, it is difficult to identify rare genetic variants that are associated with disease due to their scarcity, even among large sample sizes. In this work, we present a method for identifying associations between genetic variation and disease that considers multiple rare variants and phenotypes at the same time. By sharing information across rare variant and phenotypes, we improve our ability to identify rare variants associated with disease compared to considering a single rare variant and a single phenotype. The method can be used to identify candidate disease genes as well as genes that might represent attractive drug targets.
