By [Joehanes and Munson, GB, 2017]
Framingham Heart study: 5K samples, whole blood eQTL. QC: polymorphism-in-probe effect is likely minor.
Discovery of eQTL: step-wise regression to find independent eQTL. Found 19K independent cis-eQTL and 6K trans-eQTL. Sample size is important for the power: it scales linearly with cis-eQTL, but more with trans-eQTL. Double the size from 2500 to 5000 increase the trans-eQTL by 3-4 fold.
Validation of eQTL: (1) Internal: 75% cis and 41% trans-eQTL are validated. (2) With previous studies: 50-70% previous cis-eQTL and 30-60% trans-eQTL are replicated. The replication in the other way is low due to lower power in previous work and different sequencing platforms, etc; but still 90% of cases the directions are consistent.
Distribution of eQTL: highly enriched in transcribed regions, especially first exons and 5’ UTRs (45 fold). Modest enrichment (2-fold) of trans-eQTL in regulatory regions.
Clusters of trans-eQTL: 59 clusters with 6-200 genes. Some are due to genetic effect on cell type composition. In some clusters, found enrichment of TF motifs in promoters and miRNA targets. The majority 90% of trans-eQTL are not in any of the clusters.
GWAS analysis: with CAD/MI 58 loci, 21 loci or 36% are lead cis-eQTL. Also an example where a SNP is the trans-eQTL of a cluster of genes (SH2B3 locus).