Genome Wide Association Study

A typical GWAS analysis tests for variant significance in a set of independent samples. The most common source of sample dependence is family relationships. Yet, our study is based on a family designed cohort. For this reason, we applied a family-based test for variant significance. To perform this, we used lmer in order to create a linear mixed model for each phenotype. This model includes family id as a random effect. To correct for multiple testing, we set the significance threshold to 10^–4, which is the typical significance level a = 0.05 divided to the number of independent SNPs (497) based on linkage disequilibrium (LD) (Johnson et al., 2010; Clarke et al., 2011). We used the plink tool v.1.90 in order to calculate the LD score (––indep-pairwise 50 5 0.05) (Purcell et al., 2007). Finally, we presented the distribution of the p-values across the genome in Manhattan plots, and we tested for possible p-value inflation through Quantile–quantile (QQ) plots. For these plots, we used the GWASTools (Gogarten et al., 2012) library in R (scripts available upon request).