Association between lifetime PTSD diagnosis and ANK3 SNP genotypes was assessed using Fisher’s Exact test of association as computed by PLINK (v. 1.07, October 2009) (Purcell et al., 2007). A principal components (PC) analysis of 10,000 randomly chosen markers performed in EIGENSTRAT (Price et al., 2006) indicated no PTSD- or externalizing-associated population substructure (p >0.05 for each of the first 10 PC), and hence PC were not included in any models described below. Association between externalizing and ANK3 SNPs was examined using a regression model including sex and the SNP coded additively as predictors using PLINK’s --linear option. The significance of the association (for both Fisher’s test and the regression model) was determined by permutation based on 5,000 replicates. Unless otherwise noted, all p-values listed are uncorrected. The gene-level (358 SNP) multiple-testing corrected p-values (pcorrected) were determined using PLINK’s Max(T) procedure. Linkage disequilibrium plots were generated using Haploview (Barrett et al., 2005). Haplotype frequency estimation and haplotype-association analysis was performed using PLINK. The possibility of bias due to non-independence within couple pairs was explored using a Generalized Estimating Equation (GEE)(Liang
