Figure 5

Statistical power under different false discovery rates for four statistical methods. The four methods are general linear model (GLM), mixed linear model (MLM), compressed MLM (CMLM), and enriched compressed MLM (ECMLM). The CMLM was performed with the un-weighted pair group method with arithmetic mean clustering algorithm and the average algorithm for calculating group kinship. The ECMLM was performed by the best combination of three group kinship algorithms and eight clustering algorithms. The statistical power was evaluated on a simulated phenotype with the quantitative trait nucleotide (QTN) effect added to observed phenotypes. The size of the QTN effect is expressed in the unit of phenotypic standard deviation (SD). The x-axis indicates the false discovery rate; the y-axis shows statistical power. The observed phenotypes are (A) body mass index in human with SD = 0.08, (B) Orthopedic Foundation for Animals score in dog with SD = 0.3, and flowering time in both (C) maize with SD = 0.4 and (D) Arabidopsis with SD = 0.75.