Genotype-free individual genome reconstruction of Multiparental Population Models by RNA sequencing data

Transcriptome profiling by RNA sequencing (RNA-Seq) of genetically segregating populations is widely used to investigate the regulatory programme of gene expression and its downstream phenotypic consequences. We address analytical challenges of RNA-Seq experiments in multi-parent populations (MPPs) that are derived from two or more inbred founder strains. Genotyping arrays or low-coverage DNA sequencing are commonly used to detect founder strain variants and to reconstruct the founder haplotype mosaic of MPP individuals. We propose and evaluate a new method, Genome reconstruction by RNA-Seq (GBRS), that simultaneously reconstructs individual diploid genomes and quantifies total and allele-specific expression directly from RNA-Seq data. We demonstrate that there is sufficient information in the genetic variants revealed in RNA-Seq data to reconstruct MPP haplotypes. Unlike existing RNA-seq-based genotyping methods, GBRS exploits multi-way allele-specific expression and jointly use closely neighboring variants. GBRS haplotype reconstructions have accuracy comparable to array-based reconstructions. GBRS provides RNA-Seq quantification that is tailored to individual genomes and does not suffer from biases that can arise when using reference genome alignment. GBRS also provides a quality control for detecting sample mix-ups and can improve power to detect expression quantitative trait loci. GBRS software is freely available at https://github.com/churchill-lab/gbrs.