Two-stage study designs combining genome-wide association studies, tag single-nucleotide polymorphisms, and exome sequencing: accuracy of genetic effect estimates

Table 1 Bias in genetic effect estimates for a common tag SNP and multiple rare causal SNPs

Trait	Gene	Population	Tag SNP						Rare SNP collapsing statistic r_i/n_i
			SNP	Significance level for additive test	Estimated power for additive test	Mean effect estimate			Correlation between tag SNP and r_i/n_i	Mean effect estimate
						Over all data sets	Over data sets with significant tag effect	Relative bias (%)		Over all data sets	Over data sets with significant tag effect	Relative bias (%)
Q1	KDR	CEPH	C4S1878	0.001	0.15	0.63	0.93	49	0.41	5.51	6.19	12
DS	PIK3C2B	Tuscan	C1S9170	0.05	0.10	0.87	2.26	160	0.40	6.69	7.08	6
DS	PIK3C2B	Tuscan	C1S9171	0.05	0.10	0.87	2.26	160	0.40	6.69	7.08	6
DS	PTK2B	CEPH	C8S911	0.05	0.10	0.67	1.49	122	0.51	1.99	2.91	46
DS	PTK2B	Chinese	C8S925	0.05	0.06	0.21	1.23	476	0.47	0.34	1.53	350

Results for scenario 2. Values computed as described in the Results section. DS is disease status.

ISSN: 1753-6561