Using the same parameter values as specified in the direct scenario (Table 1), but allowing for 18 SNPs tested (per-SNP significance threshold p=0.0028) increases the required number to 2749 cases and 2749 controls in the Genetic Power Calculator. Reducing the number of independent tests based on the SNP results to 11 independent tag SNPs (per-SNP significance threshold p=0.0046), reduces the required number of cases and controls to 2531 of each. This result assumes that the disease SNP was among the tag SNPs. After allowing for a mean r2 of 0.97 between tag SNPs, we need approximately 2609 cases and 2609 controls (using a minimum r2 value of 0.8 increases this to 3164 cases and 3164 controls). Figure 1a shows the required number of cases (assuming a control:case ratio of 1:1 and an r2 correction of 0.97) for different options of disease allele frequency and GRRs, under a multiplicative disease model (GRRAA=(GRR)Aa2), that achieve 80% power of detection with SNP-based significance thresholds of p=0.0028 (18 SNPs) and p=0.0046 (11 SNPs), respectively.
