Table 1.

Estimates of variance explained by common SNPs

Phenotype	Population	Sibling	Total liability	Genotyped SNPs		Imputed SNPs
	Prevalence	Relative risk	Heritability, ${\text{h}}_{\text{T}}^2$	${\text{h}}_{{}_{\text{O}}}^2$	${\text{h}}_{\text{L}}^2$	${\text{h}}_{{}_{\text{O}}}^2$	${\text{h}}_{\text{L}}^2$	${\text{h}}_{\text{C}}^2$
All epilepsy (1258 cases)	0.005	3.3 [2.5–4.3]	32 [24–41]	31 (6)	23 (4)	42 (6)	31 (5)	26
Focal (958 cases)	0.003	2.6 [1.2–5.3]	23 [5–43]	33 (6)	27 (5)	41 (7)	33 (5)	27
Non-focal (300 cases)	0.002	4.7 [2.1–10.8]	36 [15–59]	21 (7)	38 (12)	24 (8)	46 (14)	44

For each phenotypes, we report estimates of ${\text{h}}_{{}_{\text{O}}}^2$, the percentage of variance explained on the observed scale (cases 1, controls 0), and ${\text{h}}_{\text{L}}^2$, the corresponding estimate on the liability scale (standard deviations provided in parentheses). ${\text{h}}_{\text{C}}^2$ is obtained from ${\text{h}}_{\text{L}}^2$ by subtracting the estimated inflation due to population stratification and genotyping errors (see text). For comparison, ${\text{h}}_{\text{T}}^2$, the total liability heritability, is estimated for each phenotype based on the prevalence and reported estimates of sibling relative risk (95% CI shown in square brackets).