The weights for this study were derived from the GWAS on BMI as conducted by the Genetic Investigation of Anthropometric Traits (GIANT) consortium (Speliotes et al., 2010). Details on the sample and the analysis procedures can be found elsewhere (Speliotes et al., 2010). Briefly, the GIANT consortium conducting a GWAS on BMI using 249,796 individuals and made the association signals for each of the ~2.8M SNP markers available to the public. In particular, the p-values from the GWAS served as the prior information used to devise the weighting scheme for the genome-wide association scan from this study. In the original introduction to this approach, Roeder et al. (2006) introduced exponential and cumulative weighting procedures. We opted for a cumulated weighting scheme that can be less sensitive to large prior association signals and we also used a scaling factor (B) of 2 (Roeder et al., 2006).
