Two‐Phase Designs to Follow‐Up Genome‐Wide Association Signals With DNA Resequencing Studies

Abstract

Genome-wide association studies (GWAS) of complex traits have generated many association signals for single nucleotide polymorphisms (SNPs). To understand the underlying causal genetic variant(s), focused DNA resequencing of targeted genomic regions is commonly used, yet the current cost of resequencing limits sample sizes for resequencing studies. Information from the large GWAS can be used to guide choice of samples for resequencing, such as the SNP genotypes in the targeted genomic region. Viewing the GWAS tag-SNPs as imperfect surrogates for the underlying causal variants, yet expecting that the tag-SNPs are correlated with the causal variants, a reasonable approach is a two-phase case-control design, with the GWAS serving as the first-phase and the resequencing study serving as the second-phase. Using stratified sampling based on both tag-SNP genotypes and case-control status, we explore the gains in power of a two-phase design relative to randomly sampling cases and controls for resequencing (i.e., ignoring tag-SNP genotypes). Simulation results show that stratified sampling based on both tag-SNP genotypes and case-control status is not likely to have lower power than stratified sampling based only on case-control status, and can sometimes have substantially greater power. The gain in power depends on the amount of linkage disequilibrium between the tag-SNP and causal variant alleles, as well as the effect size of the causal variant. Hence, the two-phase design provides an efficient approach to follow-up GWAS signals with DNA resequencing.

Two‐phase designs for joint quantitative‐trait‐dependent and genotype‐dependent sampling in post‐GWAS regional sequencing

“…Further extensions could also include more than one SNP genotype as stratification factors, for example, Schaid et al. ().…”

“…The fact that is a discrete covariate is important in our formulation to ensure existence of the maximum likelihood estimate (MLE) (van der Vaart & Wellner, 2001). Further extensions could also include more than one SNP genotype as stratification factors, for example, Schaid et al (2013). To operationalize the QT sampling, we discretize the QT ( ) into a three strata variable with labels (T1, T2, T3).…”

“…Under a case-control design, sampling within strata defined by both disease status and covariates is more powerful than sampling from cases and controls only (Breslow & Chatterjee, 1999;Schaid, Jenkins, Ingle, & Weinshilboum, 2013). For a quantitative trait (QT), sampling strategies that select informative individuals according to extreme-trait or genotype values are known to have good properties (Chen et al, 2012;Lin et al, 2013;Thomas, Yang, & Yang, 2013), but designs based on joint sampling are not well developed.…”

Statistical challenges in high‐dimensional molecular and genetic epidemiology

“…Once a SNP or a region with evidence of genetic association is detected, statistical fine‐mapping studies aim to acquire information on all possible causal variants in a region and evaluate relative evidence for causality (Faye et al, ; Spain & Barrett, ). Two‐phase designs present opportunities for gains in cost efficiencies in both molecular and genetic epidemiology study design, although most work to date has focused on the GWAS setting (Thomas et al, , ; Lin et al, ; Schaid et al, ). A primary motivation for two‐phase designs stems from the prohibitive cost of any emerging technology.…”

A Note on the Efficiencies of Sampling Strategies in Two‐Stage Bayesian Regional Fine Mapping of a Quantitative Trait

“…For a quantitative trait, Chen et al [] found that estimation efficiency can be gained when the frequency of sampling the homozygote strata is higher than one would expect under SRS and also when the frequency of samples from the heterozygote stratum is lower than under SRS, provided that the additive genetic model is correctly specified and the tag‐seq linkage disequilibrium (LD) is reasonably high, for example, above 0.80. In a case‐control setting, Schaid et al [] showed that stratified sampling based on both tag genotypes and case‐control status is not likely to have lower power than stratified sampling based only on case‐control status, and can sometimes have substantially greater power. Both these studies considered analysis under an additive or log‐additive model for a functional sequence variant, an assumption that may be violated in practice.…”