Statistical power for identifying nucleotide markers associated with quantitative traits in genome-wide association analysis using a mixed model

• We explore statistical power for genetic association with quantitative traits.
• We estimate empirical statistical powers from simulated data using a mixed model.
• Heritability, the number of variants and genetic variance ratio affect the power.
• Their interactions also influence the power.
• The power values help identify the degree of false negative associations in GWAS.

Use of mixed models is in the spotlight as an emerging method for genome-wide association studies (GWASs). This study investigated the statistical power for identifying nucleotide variants associated with quantitative traits using the mixed model methodology. Quantitative traits were simulated through design of heritability, the number of causal variants (NCV), the number of polygenic variants, and genetic variance ratio of causal to polygenic variants (VRCTP). Statistical power estimates were influenced not only by individual factors of heritability, NCV, and VRCTP, but also by their interactions (P < 0.05). As the genetic variance ratio decreased, the difference in power between heritabilities of 0.3 and 0.5 increased with the use of 20 causal variants, but decreased when there were 100 causal variants (P < 0.05). The power empirically estimated from the simulation study would be applicable to the design of GWAS for quantitative traits with known genetic parameters by predicting the degree of false negative associations.