Designing genome-wide association studies: sample size, power, imputation, and the choice of genotyping chip.

Schematic of how power is estimated.At the top of the figure is the recombination map and haplotypes estimated from the HapMap project [1]. Using this population genetic information we simulate a case-control sample (grey lines) where the red dots indicate the disease locus and blue dots indicate linked genetic variation. By performing a test of association at each SNP on the genotyping chip we can estimate power by counting the number of simulation for which a test statistic exceed a significance threshold (dotted line). We compare genotyping chips by changing the set of SNP at which we carry out a test. See Methods.

Plots of power (solid lines) and coverage (dotted line) for increasing sample sizes of cases and controls (x-axis).From left to right plots are given for increasing effect sizes (relative risk per allele). Both power and coverage range from 0 to 1 and are given on the y-axis. Results are for single-marker test of association and for simulations where the risk allele frequency of the causal allele is >0.05. The top row shows power for case-control studies simulated in a Caucasian population based on the CEU HapMap panel. The bottom row relates to case-control studies simulated from the YRI HapMap panel.

Power for Common versus Rare alleles.Plots of power (solid lines) and coverage (dotted line) for increasing sample sizes of cases and controls (x-axis). From left to right plots are given for increasing effect sizes (relative risk per allele). Both power and coverage range from 0 to 1 and are given on the y-axis. Results are for single-marker test of association. The top two rows show the power for rare risk alleles (RAF<0.1) and the bottom two rows show the power for common risk alleles (RAF>0.1). Rows 1 and 3 show power for case-control studies simulated in a Caucasian population based on the CEU HapMap panel. Rows 2 and 4 relate to case-control studies simulated from the YRI HapMap panel.

Histograms of the proportion of SNPs in the 22 1Mb regions (see Methods) in HapMap Phase II for which the maximum r2 with a SNP on the genotyping chip in in one of eleven bins (increasing in correlation (LD) from left to right).The same histograms are coloured in two ways. The top row shows in red the percentage of the SNPs in each bin detected (See Methods and text) when selected to be the causal SNP in our simulations (the proportion of the total volume of the bars coloured red is therefore an estimate of power). In the bottom row all r2 bins above 0.8 are coloured red (the proportion of the total volume of all the bars is therefore an estimate of coverage). Note that the use of HapMap data in choosing SNPs for the Illumina chip leads to a higher proportion of SNPs in high r2 bins.