Replication of findings in different studies can strengthen the evidence for a real association. However, this practice appears to be overrated. The majority of SNPs in the commercially available arrays are chosen to conveniently tag regions of the human genome with the consequence that very few SNP determine changes of aminoacids, and very few are located in known regulatory regions of the genome. Therefore, associations discovered through GWAS identify chromosomal regions that need finer mapping or sequencing studies to find the functional variants responsible for the disease. This intrinsic limitation of GWAS should imply that the identification of the same chromosomal region in independent studies is sufficient to move from statistical association of convenient SNPs to the discovery of the true variants or other regulatory elements that can lead to novel biological insights.
