Empirical Data Indicates a Primarily Additive Genetic Model for Expressional QTLs

Additive and dominant/recessive traits are frequently encountered in laboratory and natural populations, but the fraction of loci segregating dominant alleles has not been investigated with experimental data on many traits in the human population. Existing data and recent theoretical work suggest that the majority of genetic variation is additive, but the distribution of the fit of individual loci across the genome has not been surveyed. Here we show a smooth distribution ranging from purely additive to over-dominant/recessive genetic effects in liver gene expression traits in two independent human association study. A substantial majority of loci show additive inheritance, with a minority being dominant/recessive. Such trend holds well across minor allele frequency spectrum. Further, we looked at the most trustworthy expression quantitative trait loci (eQTLs) that were replicated in the two studies. Strikingly, only the eQTLs following additive model were replicated across studies, indicating the vast majority of true eQTLs act in additive fashion. Our results agree nicely with theoretical works and twin study evidence on clinical phenotypes. Trans-eQTLs, mediated by molecules beyond chromatin level, were also governed by additive model. Therefore, we generalize our findings to clinical trait, and argue testing for linear trend could be an efficient and powerful approach in GWAS.