A rapid and efficient linear mixed model approach using the score test and its application to GWAS

Abstract

Recent studies have demonstrated that Linear mixed model (LMM) successfully correct for population structure and genetic relatedness in genome-wide association studies (GWAS). However, currently available LMM methods usually suffer from computational inefficiency. Here, we propose a novel method, consisting of an LMM using the score test (LMM-Score). Our method takes advantage of score test that does not require estimation of parameters under the alternative hypothesis or the full model. We performed extensive simulation studies to estimate the statistical power of LMM-Score with various assumptions of sample size or marker size under different trait heritabilities. Although the limited power of simulation studies due to small sample size and conservativeness of Bonferroni correction, LMM-Score reduces the computing time greatly and maintained statistical power when compared with routine LMM method. Therefore, we believe that the LMM-Score is a valuable method for identifying the genetic basis of complex traits. We also applied our approach to real beef cattle data and suggested SOX17, RP1, LYN RPS20, snoU54, U1, MOS, PLAG1, CHCHD7, and SDR16C5 on BTA14 as the most promising candidate genes affecting pure meat weight (PMW), foreshank weight (FSW) and silverside weight (SSW) in Chinese Simmental beef cattle.