Fig. 6

Optimizing the eQTL mapping workflow increases eGene discovery. The eGene discovery power after optimizing for aggregation level, including random effect reflecting sampling variation (SV, 1/#cells), and applying cFDR methods (conditioning on matched bulk p values) at an FDR of 5% for the respective method (p < 0.05 Bonferroni). Blue bars indicate eGenes that were replicated in all bulk (a-bulk) defined as FDR < 10% and the same sign. Yellow bars indicate eGenes that were not replicated in bulk (genes tested = 20,334). * denotes multiple testing correction methods that control the (much stricter) family-wise error rate