FFselect: An improved linear mixed model for genome-wide association study in populations featuring shared environments confounded by relatedness

Linear mixed models are effective tools to identify genetic loci contributing to phenotypic variation while handling confounding due to population structure and cryptic relatedness. Recent improvements of the linear mixed model for genome-wide association analysis have been directed at more accurately modeling loci of large effect. We describe FFselect (https://github.com/NicholSchultz/FFselect), a novel method that both builds upon recent advances and further extends the linear mixed model for genome-wide association analysis to allow modeling of shared environmental effects. FFselect improves power, controls false discovery rate, and simultaneously corrects for environmental confounding to improve the utility of GWAS.