Read mismapping from segmental duplications drives spurious trans associations

Abstract

Heritable variation in gene regulation shapes both disease risk and evolution. Trans-acting quantitative trait loci (trans-QTLs) are the primary source of this variation within species, but they remain difficult to identify correctly due to false positives stemming from RNA and DNA read mismapping. Here, we identify four categories of RNA and DNA mismapping due to segmental duplications that lead to false-positive trans-QTLs. Such mismapping leaves a footprint in the form of long-range genotypic disequilibrium, which we use to identify spurious trans-QTLs. Applying this method to Genotype-Tissue Expression (GTEx) trans-QTL, we conservatively estimate that 14% of expression QTLs, 10% of splicing QTLs, and 3% of protein QTLs are artifacts. More broadly, this framework could be applied to flag spurious distant peaks in genome-wide association studies, Hi-C contacts, and co-expression between distant genes.