Accurate Resolution of Gene and Isoform Allele-Specific Expression in Human Liver and Brain

Genetic variation modulates gene expression transcriptionally or post-transcriptionally, and can profoundly alter phenotype. Measuring differential expression at heterozygous loci between alleles within an individual, a phenomenon called allele-specific expression (ASE), can assist in identifying such factors. Massively parallel DNA and RNA sequencing and advances in bioinformatic methodologies provide an outstanding opportunity to measure ASE genome-wide. In this study, matched DNA and RNA sequencing, genotyping arrays and computationally phased haplotypes were integrated to systematically evaluate ASE in human brain and liver tissue. We describe a comprehensive methodology to accurately measure SNP, gene and isoform ASE through the use personalized haplotype genome alignment, comprehensive alignment quality control and intragenic SNP linking. Our results indicate that accurate ASE quantification requires careful bioinformatic analysis and is adversely affected by random sampling even at moderate sequence depths. We identified multiple known and several novel ASE genes including WDR72, DSP and UBD, however the majority of intragenic ASE SNPs we observed showed imbalance discordant with haplotype phase, many of which we explain by observed transcript structure, suggesting that much genic ASE is isoform derived. The methods used in this study will be of use to researchers quantifying ASE, and the genes and isoforms identified as ASE of interest to researchers studying those loci.