Table 3.

Required Total Sample Size to Achieve 80% Power to Detect Rare Variants Associated with a Continuous or Dichotomous Case-Control Phenotype at the Genome-wide Level α = 10⁻⁶

Total Sample Size	Maximum β = 1.6/ Maximum OR = 5		Maximum β = 1.9/ Maximum OR = 7
	5% Causal	10% Causal	5% Causal	10% Causal
Continuous trait	5,990	1,800	4,260	1,290
Dichotomous trait with prevalence 10%	15,120	4,810	9,650	3,120
Dichotomous trait with prevalence 1%	12,030	3,870	7,010	2,290

Power was estimated via the analytical formulae assuming 5% or 10% of variants with MAF < 3% are causal. Regression coefficients for the s causal variants were assumed to be a decreasing function of MAF, $\left|{\beta }_j\right|=c{|\log }_{10}MA{F}_j|$(j = 1,…,s), where 80% of β_j's are positive and 20% are negative; see Figure S2. Required total sample sizes (cases and controls) are given for different “maximum” effect sizes (or ORs) when MAF = 10⁻⁴ and different prevalences for case-control studies. Estimated sample sizes were averaged over 100 random 30 kb regions.