For assessing cross-tissue replication, we used a large whole-blood eQTL dataset from the DGN study26 and two smaller eQTL datasets from the Immune Variation (ImmVar) study27 that consist of monocyte and T cell data. π1 of these eQTLs in our dataset are 0.63 (whole blood), 0.61 (monocytes), and 0.67 (T cells), which are greater than their empirical null mean of 0.10 (p < 0.0001 for all three datasets, one-tailed). Thus, a large proportion of blood eQTLs are present in our brain data. We also assessed the replication rate of our brain-derived eQTLs in the whole-blood DGN dataset (Figure 2A–B). When we consider SNP-gene pairs that can be tested in both studies, we observed a replication rate of 0.83 (Figure 2C), which is greater than its empirical null mean of 0.30 (p < 0.0001, one-tailed). This increase in replication rate may be due to the higher statistical power of the DGN study and the fact that cortical tissue consists of a large variety of cell types, which in aggregate, expresses a large proportion of the transcriptome. Since blood contains a mixture
