For each sample, allele-specific RNA-seq read counts were generated at all heterozygous variants with the GATK ASEReadCounter tool75. Only uniquely mapping reads with a base quality ≥10 at the variant were counted, and only those variants with coverage of at least eight reads were reported. Variants that met any of the following criteria were flagged and removed from downstream analyses: 1) UCSC 50-mer mappability of <1; 2) simulation-based evidence of mapping bias76; and 3) heterozygous genotype not supported by RNA-seq data across all samples for that donor and no significant (FDR > 1%) evidence that the variant is monoallelic in expression data75. Gene level measurements of haplotype expression were calculated by aggregating counts per sample across all heterozygous variants with ASE data within the gene using population phasing. Full ASE data are available through dbGaP.
