Skip to main content
. 2021 Sep 4;19(4):629–640. doi: 10.1016/j.gpb.2021.08.005

Supplementary Figure S3.

Supplementary Figure S3

Comparison between single-locus and multi-locus methods on power against FDR and type I error. Single-locus methods include GLM and MLM, whereas the multi-locus methods include FarmCPU and BLINK. The comparison was based on a simulated trait using the maize data containing 282 individuals and 3094 SNPs (GAPIT demonstration datasets, https://zzlab.net/GAPIT/GAPIT_Tutorial_Data.zip). The simulated trait had a heritability of 75% controlled by 20 QTNs. Power was calculated as the proportion of QTN detected. FDR was calculated as the proportion of non-QTNs among the positives (A). Type I error was calculated as the proportion of tests with false positives (B). The simulation was replicated 100 times. FDR, false discover rate.