To identify associations with rare variants, just WHaIT: Weighted haplotype and imputation-based tests

Abstract

Empirical evidences suggest that both common and rare variants contribute to complex disease etiology. Although the effects of common variants have been thoroughly assessed in recent genome-wide association studies (GWAS), our knowledge of the impact of rare variants on complex diseases remains limited. A number of methods have been proposed to test for rare variant association in sequencing-based studies, a study design that is becoming popular but is still not economically feasible. On the contrary, few (if any) methods exist to detect rare variants in GWAS data, the data we have collected on thousands of individuals. Here we propose two methods, a weighted haplotype-based approach and an imputation-based approach, to test for the effect of rare variants with GWAS data. Both methods can incorporate external sequencing data when available. We evaluated our methods and compared them with methods proposed in the sequencing setting through extensive simulations. Our methods clearly show enhanced statistical power over existing methods for a wide range of population-attributable risk, percentage of disease-contributing rare variants, and proportion of rare alleles working in different directions. We also applied our methods to the IFIH1 region for the type 1 diabetes GWAS data collected by the Wellcome Trust Case-Control Consortium. Our methods yield p values in the order of 10⁻³, whereas the most significant p value from the existing methods is greater than 0.17. We thus demonstrate that the evaluation of rare variants with GWAS data is possible, particularly when public sequencing data are incorporated.

Figure 1

Comparison of Power by r, Percent of Rare Alleles in the Causal Region that Increase Disease Risk Power of all tests was assessed at the 5% level by using empirical significance threshold determined by 1000 null data sets per region. 50% of the rare alleles in the causal region were assumed to contribute to disease risk (i.e., d fixed at 50%), and the PAR of each contributing SNP was fixed at 0.5%.

Figure 2

Comparison of Power by PAR Power of all tests was assessed at the 5% level by using empirical significance threshold determined by 1000 null data sets per region. 50% of the rare alleles in the causal region were assumed to contribute to disease risk (i.e., d fixed at 50%), and all contributing rare alleles were assumed to increase disease risk (i.e., r fixed at 100%).

Figure 3

Comparison of Power by d, Percent of Disease-Contributing Rare Variants Power of all tests was assessed at the 5% level by using empirical significance threshold determined by 1000 null data sets per region. All rare alleles in the causal region were assumed to increase disease risk (i.e., r fixed at 100%), and the PAR of each contributing SNP was fixed at 0.5%.