Extended Data Fig. 6: External validation of eQTL.
From: Genetic regulation of gene expression across multiple tissues in chickens
(a) linear regression model (LRM)-based validation. (b) linear mixed model (LMM)-based validation. Validation was carried out in three tissues: hypothalamus (upper row), liver (middle row) and pituitary gland (bottom row). ChickenGTEx served as the discovery population, while an independent validation population consisted of commercial White Plymouth Rock chickens. P values were computed via the asymptotic t approximation. (c) Comparison of SNPs between the ChickenGTEx discovery population and validation populations. Only SNPs common (minor allele frequency, MAF > 0.05) in both discovery and validation populations were included. (d) The number of SNPs used in π1 calculation shown in panel (a). (e) Sample sizes of the ChickenGTEx discovery and validation populations. (f) The number of eGenes detected in the discovery and validation populations. (g) Effect size distribution of replicated and not-replicated eQTLs across tissues. Replicated eQTLs: SNPs significant in the discovery population that also meet the significant threshold in the validation population. Boxplot details: The central band represents the median, the box boundaries represent the interquartile range (25% to 75%), and the whiskers extend 1.5 × the interquartile range. Statistical significance was obtained using a two-sided Student’s t-test. (h) The π1 value plotted as a function of eQTL effect size (log2 transformed allelic fold change, log2aFC) in the discovery population (liver). (i) Histogram of eQTL nominal P values in the validation population. The nearest variant in the validation population to the corresponding lead variant in the discovery population was selected for each eGene. (j) Distribution of eQTL nominal P values in the validation population. For each eGene identified in the discovery population, the top lead variant within the same LD block was selected.
