low risk genotypes. We note that although this second replication condition is heuristic, (1) it is a higher dimensional analog for the routine procedure of checking univariate results to ensure that allelic effect is in the same direction in the initial and replication datasets, and (2) it provides an additional way to reject replication and cannot increase the number of false positives. Because only a limited number of the COGEND SNPs were genotyped in the ACS data, this testing for replication was only available for a few signals identified in the COGEND data. For one of the COGEND SNPs not genotyped in the ACS data, rs3743075, linkage disequilibrium information from both COGEND and HapMap suggests that rs514743 is an appropriate surrogate (r2 = 0.97 in COGEND data, r2 = 0.92 in the HapMap release 22 CEU data). The use of rs514743 as a proxy for rs3743075 allowed us to evaluate one additional signal from the COGEND data in the ACS data.
