We report two SNPs highly significant in our discovery data for the phenotypes CCT and CDR. Both have positive replication data, neither of which withstands a multiple testing correction, but both show direction of effects consistent with the discovery data. The most significant SNP in the POAG risk analysis is located in NEDD9. There was strong evidence from surrounding SNPs and NEDD9 appears an excellent candidate as it has been shown to be increased in trabecular meshwork cells [23], however, none of the 3 SNPs chosen for follow-up in the NEDD9 region were significant in the replication study. As GWAS studies are prone to type one (false positive) error, we chose to follow-up the SNPs with most evidence, based on a variety of criteria, rather than simply single p-values. We believe our selection criteria enhanced our likelihood of successful replication, but it is possible we may have excluded some real associations.
