Probing the limits of cis-acting gene regulation using a model of allelic imbalance quantitative trait loci

Imbalance in gene expression between alleles is a hallmark of cis-acting expression quantitative trait loci (eQTLs) and several methods have been developed to exploit allelic imbalance to support the identification of eQTLs. Allelic imbalance is also of scientific and, potentially, clinical interest as it can erode the degree to which the effects of deleterious variants are buffered in a diploid organism and has been reported to be associated with the penetrance of pathological genomic variants. Here, we develop and apply a statistical model that is designed to evaluate whether the genotype of a locus is associated with the degree of allelic imbalance of a gene and refer to such loci as allelic imbalance quantitative trait loci (aiQTLs). An advantage of our approach is that it does not depend on linkage disequilibrium between the aiQTL and the associated gene and is, therefore, suited to the identification of eQTLs that act in cis over very large distances. We applied our model to data from the GTEx consortium and examined the relationship between the distance of an eQTL from the TSS of the associated gene and the evidence that the eQTL acts in cis. Previous studies have used a distance of 1Mb from the target gene as an indication that an eQTL acts in cis; however, our results suggest that the majority of eQTLs at distances more than 500 kb from the TSS of the target gene are likely to act in trans (and thus to affect both gene copies). The model used here is also well suited to comparing the overall extent of allelic imbalance between samples. We show that in some tissues allelic imbalance is correlated with age; however, this correlation may be due to changes in the abundance of immune cell populations with age, as we found strong correlations between sample-level allelic imbalance and the inferred abundance of multiple immune cell types across whole blood samples.