Review and further developments in statistical corrections for Winner’s Curse in genetic association studies

Genome-wide association studies (GWAS) are commonly used to identify genomic variants that are associated with complex traits, and estimate the magnitude of this association for each variant. However, it has been widely observed that the association estimates of variants tend to be lower in a replication study than in the study that discovered those associations. A phenomenon known as Winner’s Curse is responsible for this upward bias present in association estimates of significant variants in the discovery study. We review existing Winner’s Curse correction methods which require only GWAS summary statistics in order to make adjustments. In addition, we propose modifications to improve existing methods and propose a novel approach which uses the parametric bootstrap. We evaluate and compare methods, first using a wide variety of simulated data sets and then, using real data sets for three different traits. The metric, estimated mean squared error (MSE) over significant SNPs, was primarily used for method assessment. Our results indicate that widely used conditional likelihood based methods tend to perform poorly. The other considered methods behave much more similarly, with our proposed bootstrap method demonstrating very competitive performance. To complement this review, we have developed an R package, ‘winnerscurse’ which can be used to implement these various Winner’s Curse adjustment methods to GWAS summary statistics.Author summary: A genome-wide association study is designed to analyse many common genetic variants in thousands of samples and identify which variants are associated with a trait of interest. It provides estimates of association strength for each variant and variants are classified as associated if their test statistics obtained in the study pass a chosen significance threshold. However, due to a phenomenon known as Winner’s Curse, the association estimates of these significant variants tend to be upward biased and greater in magnitude than their true values. Naturally, this bias has adverse consequences for downstream statistical techniques which use these estimates. In this paper, we look at current methods which have been designed to combat Winner’s Curse and propose modifications to these methods in order to improve performance. Using a wide variety of simulated data sets as well as real data, we perform a thorough evaluation of these methods. We use a metric which allows us to identify which methods, on average, produce adjusted estimates for significant variants that are closest to the true values. To accompany our work, we have created an R package, ‘winnerscurse’, which allows users to easily apply Winner’s Curse correction methods to their data sets.