Table 8.

Genome-Wide Association Study Results for Baseline Low-Density Lipoprotein Cholesterol Using an Unrelated Subset of Subjects from the Metabolic Syndrome in Men Cohort

	LR	LMM	BOLT-INF	BOLT-LMM	cvBLUP
lambdaGC	1.02	1.02	1.02	1.02	1.02
p-values below 1e-06	31	32	32	35	33
Mean ratio of effect size estimates, 1e-06	0.991	1	0.999	0.999	0.998
Standard error in ratio of effect sizes, 1e-06	0.00243	0	0.000561	0.000561	0.00088
p-values below 5e-08	15	15	15	16	16
Mean ratio of effect size estimates, 5e-08	0.996	1	0.998	0.998	0.999
Standard error in ratio of effect sizes, 5e-08	0.00232	0	0.000977	0.000977	0.00122

LMM with GLS analysis of SNP effects implemented in GCTA; cvBLUP, cross-validated prediction-adjusted linear regression; BOLT-INF; BOLT assuming infinitesimal genetic model; BOLT-LMM, BOLT using mixture of normal distributions as prior for SNP effect sizes, that is, sparse genetic architecture. cvBLUP-adjusted analyses, LMM, and BOLT were used in a leave-one-chromosome-out scheme with variance components, cvBLUPs, covariance models (LMM, GCTA), and genetic predictions and residuals (BOLT) generated using SNPs on chromosomes other than that of the test-SNPs.

GCTA, genome-wide complex trait analysis; GWAS, genome-wide association studies; LR, linear regression; METSIM, metabolic syndrome in men.