Graphical pangenomics-enabled characterisation of structural variant impact on gene expression in Brassica napus

Structural variants (SVs, eg. insertions and deletions) are genomic variations > 50 bp that are known to be associated with a range of crop traits, from yield to flowering behaviour and stress responses. Recently, pangenome graphs have emerged as a powerful framework for analysing genomic data by encoding population- or species-level diversity in one data structure. Pangenome graphs have the potential to serve as unbiased references for downstream applications, including SV genotyping and pan-transcriptomic analyses. In this work, we hypothesized that extensive variation affects transcript quantification and expression quantitative trait locus (eQTL) analysis when relying on a single reference, and that using pangenome graphs can mitigate reference sequence bias. We combined long and short read whole genome sequencing data with expression profiling of Brassica napus (oilseed rape) to assess the impact of SVs on gene expression regulation and explored the utility of pangenome graphs for eQTL analysis. We demonstrate that pangenome graphs provides a superior framework for eQTL analysis by eliminating single reference bias in gene expression quantification. Combined with the graph-based genotyping of SVs, we identified 240 eQTL-SVs found in close proximity of target loci. These SVs affect expression of genes related to important traits, are often not in linkage with SNPs and represent diversity unaccounted for in classical SNP-based analyses. This study highlights the multiple advantages of graph-based approaches in population-scale studies and provides novel insight into gene expression regulation in an important crop.