Sequencing & Pangenomes

The Bovine Long Read Consortium: developing a long-read sequence population to enable accurate imputation of structural variants.

---

Abstract

Structural variants (SV) are DNA mutations encompassing ≥ 50 base pairs. SV are of interest for livestock studies because, due to their large size they are more likely to have a high impact on traits of interest compared to SNP. Until recently there were no genome-wide SV association studies in large cattle populations because SV are difficult to characterise using short-read sequence technology. Recent advances in long-read sequencing now enable accurate genome-wide characterisation of SV, presenting a new opportunity to develop an SV imputation reference database for cattle. The Bovine Long Read Consortium (BovLRC) has been initiated with twenty collaborating institutes, enabling joint-calling of SV in over 700 cattle with long-read sequence and representing 50 different breeds and crosses. The BovLRC has three objectives: discovery of SV at population-scale, development of an SV imputation reference and evaluation of the impact of SV on traits of interest. It is therefore essential that both SV and SNP are accurately discovered and genotyped to maximise imputation accuracy. Unfortunately, there are no currently established bovine SV and SNP truth sets that can be used to evaluate sequence genotyping accuracy. Therefore, prior to processing the full BovLRC dataset we validated SV genotyping and imputation accuracy using a subset of 159 animals that included 22 parent-offspring duos. Using this population level genotype data, we identified 0.5 Mb genome regions that exhibited excess heterozygosity or above average genotype missingness rates. These "complex" regions included a part of the major histocompatibility complex (MHC; involved in disease immunity) and tended to show higher opposing homozygous errors in the duos compared to the remaining "reliable" regions. The autosomal SV discovered in the 159 long-read sequences (insertions and deletions only) were used to call SV genotypes in a set of 514 animals with short-read sequence to generate an SV-SNP imputation reference population. Only half of these SV passed quality filtering when called in short-read sequence due to the known limitations of short-read data. However, the imputation accuracy for the remaining SV was high (correlation between imputed and real genotypes = 0.92) in reliable regions of the genome but as expected was lower in complex regions (0.71). The SV were imputed into a 68,774 Holstein cows with stature records. Using this data, we undertook a genome-wide association study for stature and identified a very significant association with a previously reported putative causal SV that deletes a 6.2 kb region of chromosome 11, including part of the MATN3 gene. This study demonstrates the feasibility and power of accurate SV discovery, imputation and downstream analysis, while also highlighting the need for further research in complex genome regions. Next, it is important to test imputation accuracy using the complete set of BovLRC sequences.

Keywords: 2026

How to Cite:

Chamberlain, A., Dorji, J., Goddard, M., Hayes, B., MacLeod, I., Nguyen, T., Pryce, J., Reich, C., Tran, H., Vander Jagt, C., Wang, J. & Xiang, R., (2026) “The Bovine Long Read Consortium: developing a long-read sequence population to enable accurate imputation of structural variants.”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2286392. doi: https://doi.org/10.31274/wcgalp.23977

Rights: 1