PhAT-QTL: A phase-aware test for QTL detection

Abstract

Next generation sequencing based molecular assays have enabled unprecedented opportunities to quantitatively measure genome function. When combined with dense genetic data, quantitative trait loci (QTL) mapping of molecular traits is a fundamental tool for understanding the genetic basis of gene regulation. However, standard computational approaches for QTL mapping ignore the diploid nature of human genomes, testing for association between genotype and the total counts of sequencing reads mapping to both alleles at each genomic feature. In this work, we develop a new phase-aware test for QTL analysis (PhAT-QTL) leveraging the inherent single nucleotide resolution of sequencing reads to associate the alleles of each marker with the allele-specific counts (ASC) at a genomic feature. Through simulations, we show PhAT-QTL achieves increased power relative to standard genotype-based tests as a function of the number of heterozygotes for a given marker, the noise correlation between haplotypes, and the number of samples with detectable allele-specific counts at a genomic feature. Simulations further show that phasing error and error in quantifying ASC results in a loss of power as opposed to bias. Read simulations on varying haplotype structures (simulated from 1000 Genomes phased genomes) demonstrate that PhAT-QTL is able to detect 20% more QTLs while maintaining the same false positive rate as previous approaches. Applied to RNA-sequencing data, PhAT-QTL achieves similar performance as previous phase-aware methods in detecting cis expression QTLs (cis-eQTLs) but at a fraction of the computational cost.