Fig. 2.

Examples of PICALO optimization for ieQTLs. A ieQTL for the genes TUBB2A and C9orf78 with expression PC6 before optimization and resulting PIC5 after optimization in blood. B ieQTL for the genes FAM221A and ADAMTS18 with expression PC6 before optimization and resulting PIC2 after optimization in the brain. C The number of eQTLs tested in the blood and brain and the respective number of eQTLs that have an interaction with one or more PICs or expression PCs. D The number of ieQTLs for the first five PICs in the blood. E The number of ieQTLs for the first five PICs in the brain. F Regression plot showing the correlation between PIC1 and estimated RNA-seq sample quality calculated as the per-sample expression correlation with the overall average expression. G Pearson correlation heatmaps correlating PIC (top) and expression PC (bottom) to RNA-seq alignment metrics in the blood. The correlation p-values are corrected for multiple testing with Benjamini-Hochberg, and only correlations with an FDR < 0.05 are shown. Note that many of the expression PCs correlate significantly with RNA-seq alignment metrics while only a limited number of PICs show a significant correlation