It is important to note several limitations of the present study and the measures we have taken to account for them. First, our cohort is small, containing only seven OCD subjects and eight unaffected comparison subjects. Obtaining post-mortem tissue with sufficiently high RNA quality to perform robust RNA-sequencing (Table 1; Fig. S1) remains a challenge, and because of this it is difficult to evaluate the effect of all potential covariates on gene expression. Here we built a model including only covariates that statistically affected transcript expression (see Methods). A second related point is that DEG may be detected due to several factors, including (1) large OCD versus unaffected subject differences over all three regions, (2) large differences within only one region, (3) or modest but consistent differences for all three regions, among other scenarios. Global analysis could detect DEG for any of these scenarios, within the limits of statistical power, whereas regional analysis would likely detect DEG only for (1) or (2). If (3) would be a better model for most DE genes, global analysis has better power. Our small
