With these arguments in mind, we therefore consider estimation of a frequentist P–value threshold that accounts for the multiplicity of the whole genome and can be used as a baseline for calibrating other approaches. We consider two approaches to estimating this threshold, accounting for the correlation within the genome: one based on a permutation scheme, and the other using linear algebra to estimate an effective number of tests directly from genotype data [Patterson et al., 2006]. We apply both methods to genotypes released by the WTCCC, extrapolating the results to complete saturation to obtain a genomewide threshold. We conclude by discussing some implications of these results in the light of the WTCCC and other studies.
