. 2020 Apr 8;27(4):599–612. doi: 10.1089/cmb.2019.0325

Table 2.

Simulations with Infinitesimal Genetic Model, Without Population or Family Structure

Model	mean( $β_{g 0}$ )	se( $β_{g 0}$ )	mean( $χ_{N u l l}^{2}$ )	se( $χ_{N u l l}^{2}$ )	mean( $β_{g}$ )	se( $β_{g}$ )	mean( $χ^{2}$ )	se( $χ^{2}$ )
LR	−4.00E-04	1.30E-03	0.99	0.05	0.250	1.30E-03	47.11	0.66
LR + 4 PCs	−5.00E-04	1.30E-03	0.99	0.05	0.250	1.30E-03	46.97	0.66
LR + cvBLUP	−1.30E-03	1.10E-03	1.00	0.05	0.249	1.20E-03	58.53	0.81
LR + BLUP	−1.00E-03	6.00E-04	1.05	0.05	0.096	6.00E-04	36.63	0.56
LMM	−1.30E-03	1.10E-03	1.00	0.05	0.249	1.20E-03	56.62	0.76
BOLT-INF	−1.30E-03	1.00E-03	1.01	0.05	0.219	1.00E-03	56.13	0.76
BOLT-LMM	−1.30E-03	1.00E-03	1.00	0.05	0.219	1.00E-03	56.04	0.75
LR + true genetic effect	−1.00E-03	9.00E-04	1.04	0.05	0.250	9.00E-04	96.28	1.26

Mean and standard error for effect estimates and test statistics of association tests at 1000 null SNPs with true $β_{g 0} : \hat{β_{g 0}}$ and $χ_{N u l l}^{2}$ ; and at 1000 causal SNPs with an alternative hypothesis of true $β_{g} = 0.25$ : ${\hat{β}}_{g}$ and $χ^{2}$ .

LMM, linear mixed model; PC, genetic principal component; BLUP, Best Linear Unbiased Predictor; cvBLUP, cross-validated Best Linear Unbiased Predictor.