Sequencing & Pangenomes

Integrating structural variants into sequence-based association studies in dairy cattle using a pangenome and imputation framework

---

Abstract

Structural variants (SVs) are a major contributor to genetic diversity, yet they are most effectively detected using long-read (LR) sequencing. However, LR sequencing is still limited, and it often lacks associated phenotypes. To address this issue, we employed a combination of pangenome-based (variation graph) and imputation approaches to facilitate large-scale SV association studies in three French dairy cattle breeds. We constructed a variation graph from long-read-detected structural variants (SVs), including 69,892 deletions, 89,900 insertions, and 17,402 duplications identified in 176 samples. Short-read (SR) sequences from 1,280 individuals were then aligned to the graph to genotype the SVs. Validation analyses revealed high genotype concordance for deletions and insertions (both 0.79) but low concordance for duplications (0.14), resulting in the exclusion of duplications.Of the 1,280 genotyped individuals, 939 were selected as the imputation reference. Prior to imputation, a total of 141,832 SNPs falling within deletions were removed from the reference panel. Benchmarking of imputation tools showed that Beagle performed better than Glimpse and Minimac for SV imputation. Using a two-step imputation strategy, we imputed SNPs and SVs for 11,902 Holstein, 3,753 Montbéliarde, and 3,053 Normande bulls with daughter yield deviations for 13 traits related to milk production, udder health, fertility, and stature. The first step was within-breed imputation from low-density (50K) to high-density (700K) SNPs using FImpute. This was followed by imputation to the sequence level (20M variants including SVs), using Beagle. After filtering (minor allele frequency > 0.01 and imputation R² > 0.6), approximately 14 million SNPs and 40,000 SVs were retained for genome-wide association analyses (GWAS) within each breed using a linear mixed model. GWAS results were consistent with previous findings and identified thirty-six unique, significant SV-trait associations (-log10(P) > 8.4). While SNPs generally exhibited stronger signals within the loci of interest, likely due to their higher imputation accuracy (median R² = 0.99 for SNPs versus 0.70 for SVs), a 1217 bp insertion was identified as the most significant variant associated with a QTL linked to stature in Montbéliarde cattle. This SV is located within an intron of the GABRG3 gene, which is involved in growth and developmental disorders in human. In total 10 strong candidate SV were identified, annotated and discussed. Overall, our results demonstrate that incorporating structural variants (SVs) into genome-wide association studies (GWAS) provides new insights into the genetic architecture of complex traits in cattle. Our results also highlight the need to expand LR sequencing resources to improve the accuracy of SV imputation and increase the power of future association studies. This work was carried out as part of the CASCAD project, which was funded by CARNOT France Futur Élevage (F2E). The sequence data were produced by the FEDER SeqOccIn and H2020 RUMIGEN projects.

Keywords: 2026

How to Cite:

Boichard, D., Boussaha, M., Faraut, T., Fritz, S., GROHS, C., Klopp, C., Naji, M., Sanchez, M. & Sorin, V., (2026) “Integrating structural variants into sequence-based association studies in dairy cattle using a pangenome and imputation framework”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2286025. doi: https://doi.org/10.31274/wcgalp.23840

Rights: 1