But this is not what happens in practice. There is a strong consensus in the field that the results of association studies should not be relied upon without additional (statistically significant) evidence from analyses in independent replication samples [17] , and many major journals have policies which preclude publication of GWAS studies by themselves, without such replication evidence. Common practice is thus to regard the GWAS as an experiment to highlight SNPs of interest, and then to take as many as possible of the interesting SNPs into replication studies.
