A final step in selecting SNPs is, if possible, to check whether there is any evidence that genotyping is at risk of failing. For instance, it is normally considered difficult to genotype or sequence in genomic regions of low complexity, containing numerous repetitive elements. Since the genotyping could fail for some markers in these regions, it is useful to incorporate additional SNPs in order to capture local variation. This is particularly important in a replication study, where the ability to replicate hinges on the genotyping accuracy of the SNP in question. In this situation, it is advisable to include the next most highly associated SNP in high LD (r2>0.7) and/or a SNP in complete LD with the initial hit using data from the original study or HapMap information if samples are of similar ancestry.
