region containing the PTSD-associated SNPs. When we estimated the haplotype frequency, it became clear that the PTSD-associated SNPs tag a single haplotype. Starting with the 3 SNPs in perfect LD, and adding adjacent SNPs that either increased the association or did not break down association, we identified a core PTSD-associated haplotype spanning 41 kb—from rs72818458 to rs7098848 (chr10:61,799,888- 61,841,144). In particular, the AGAAAAG haplotype of SNPs 5 to 11 (Table 1) occurred at an estimated 5.1% frequency in cases and 11% in controls (7.4% frequency overall) and was more strongly associated with protection from PTSD than any individual SNP (OR=0.426, p=0.00042). The addition of other intervening SNPs to the haplotype did little to change the frequency of the haplotype defined by SNPs 5 to 11 or alter the evidence of association with PTSD. This reinforces the notion that these SNPs tag a single ancestral haplotype that is likely to harbor a protective allele. Additionally, when 79 subjects predicted to have the protective haplotype with probability ≥ 50% were removed from the analysis, SNPs 5 to 12 in Table 1 were no longer associated with PTSD (all p> .50, OR between 0.8 and 1.1 based on an analysis of 267 cases
