Referent cohort subjects used in other studies can also be used as controls. This was the strategy used by the Wellcome Trust Case–Control Consortium that used the same pool of 3,000 controls chosen from the British population to search for genetic modifiers of seven common traits [6]. This strategy is becoming more and more feasible with the increasing availability of GWAS data from dbGaP, the database of genotype–phenotype associations (http://www.ncbi.nlm.nih.gov/gap), and the Illumina control database. However, this approach can introduce confounding due to population stratification. Population stratification has been a well known source of spurious associations [48] and occurs when allele frequency differences in cases and controls are due to differences in ancestry rather than associations between genes and disease. Population stratification is, however, easy to detect in GWAS using multivariate statistical techniques implemented in popular programs such as PLINK [51] and EIGENSOFT [52]. Figure 2 provides an example using data from the New England Centenarian Study [53] and the Human Genome Diversity Panel [54]. The analysis can help to describe the genetic background of cases and controls when reference
