An Exhaustive Scan Method for SNP Main Effects and SNP × SNP Interactions Over Highly Homozygous Genomes

Genome-wide association studies (GWAS) have been a powerful tool for exploring potential relationships between single-nucleotide polymorphisms (SNPs) and biological traits. For screening out important genetic variants, it is desired to perform an exhaustive scan over a whole genome. However, this is usually a challenging and daunting task in computation, due mainly to the large number of SNPs in GWAS. In this article, we propose a computationally effective algorithm for highly homozygous genomes. Pseudo standard error (PSE) is known to be a highly efficient and robust estimator for the standard deviation of a quantitative trait. We thus develop a statistical testing procedure for determining significant SNP main effects and SNP × SNP interactions associated with a quantitative trait based on PSE. A simulation study is first conducted to evaluate its empirical size and power. It is shown that the proposed PSE-based method can generally maintain the empirical size sufficiently close to the nominal significance level. However, the power investigation indicates that the PSE-based method might lack power in identifying significant effects for low-frequency variants if their true effect sizes are not large enough. A software is provided for implementing the proposed algorithm and its computational efficiency is evaluated through another simulation study. An exhaustive scan is usually done within a very reasonable runtime and a rice genome data set is analyzed by the software.