Our simulations also assume that all rare variants identified through re-sequencing of the discovery samples will subsequently be genotyped in the analysis cohort. However, genotyping these rare variants on a genome-wide scale will be a considerably more expensive endeavour than utilising GWA platforms. This approach may currently be financially infeasible with the large samples required to detect the modest genetic effects we expect for complex traits, particularly for rare variant associations. One possible approach to reduce genotyping costs is to focus on potentially functional rare variants (for example those leading to non-synonymous changes or located within exons and regulatory regions). However, at present, there is no unbiased evidence to suggest that causal variants are more likely to aggregate in such regions and current annotation of the genome is incomplete, making identification of potentially functional loci difficult.
