Next, we examined the extent to which PC1 appeared to dominate the case-control comparisons between participants with alcohol dependence and controls. To accomplish this, we repeated the case-control comparisons having first removed PC1 from the data. Specifically, PCA (as described above) was performed on residualized data of all participants including cases and controls. After obtaining PC loadings, individual data were reconstructed using all PCs but excluding PC1. In this way, variances associated with PC1 were removed from the data. This differs from dividing regional CT by mean CT or from adjusting mean CT in the group comparison wherein regions are weighted equally. Then group comparisons were performed. Note that the original case-control comparison was performed on the raw data while controlling nuisance covariates. To ensure residualization did not interfere with the comparison between the two analytical strategies, the standard case-control comparison in this exploratory analysis was performed on the residualized data.
