retinol [38] and e-selectin [39]. However, GWAS of secondary traits face practical issues in terms of different genotyping arrays, low variability in the phenotype of interest within a single GWAS (e.g. rare diseases where only a handful of cases may occur in the original GWAS), and theoretical issues including ascertainment bias due to oversampling of cases [40] or differential genotype/imputation quality between studies [41] (e.g. if controls are “utilized” from GWAS data generated on a different genotype platform).
