Immediately apparent is how close, for studies in Caucasian populations, the genotyping chips track the power afforded by the ideal “Complete chip” in a given study design and disease model. Figure 3 illustrates that the potential benefits of increasing SNP density on the chips or from using imputation [14] are greatest for low frequency SNPs. When focusing on common alleles, the potential benefits are greatest for the Affymetrix 100 k and 500 k chips and the Illumina 300 k chip and we show this when specifically consider imputation below (see Table 1). However, a clear consequence of these results is that for any of the chips in current use, increasing sample size is likely to have a bigger effect on power than increasing SNP density.
