Figure 1

Overview of integrative pyramid analysis schema for the genome-wide discovery. Each stage (represented by an edge of the pyramid) represents a model describing the relationship between the two variables (represented by the nodes for the edge). Each stage uses only the data inherited from the previous stages which may have utilized certain thresholds for inclusion criteria. Thus, the filtering stages (stages 1-5) form a series of biologically relevant steps that result in a reduction in the number of SNP tested at the genetic association testing stage (stage 6) and therefore a potential increase in power. A permutation procedure (see Methods) that accounts for the fact that the samples for the filtering stages may overlap with the samples used at the genetic association testing stage was used to evaluate the overall significance of the resulting SNPs. For each replicate derived from the application of the pyramid analysis to a permuted dataset (in our analysis, n = 1000), we obtain a list of SNPs and p-values for association with the trait (designated here as p ₁ , p ₂ , p ₃ , etc.). We utilize these sets of p-values derived from permuted datasets to estimate the null distribution.