A deterministic algorithm for optimality of threshold in a GWAS experiment

While genome-wide association study (GWAS) is an important tool for gene discovery for economic traits in livestock, its use of large numbers of genetic markers necessitates the use of multiple testing correction methods. Several of these methods have been suggested, but their optimality is not as well studied. The aim of this study is to present a deterministic algorithm to provide a framework for estimating the power and false positive rate (FPR) in a GWAS, and using these estimates to test the optimality of these correction method based on the Receiver Operating Characteristic (ROC) curve. This study suggests that both Bonferroni correction and Benjamini-Hochberg False Discovery Rate are overly conservative even if under the assumption of independence between markers.