Sequencing & Pangenomes

Accurate structural variation imputation using low-pass sequencing data in a turkey population

---

Abstract

Structural variants (SVs) are an important source of genetic diversity, yet their accurate genotyping in large populations remains challenging, particularly when relying on array based genotyping. Here, we present an approach to impute SVs in a turkey (Meleagris gallopavo) population using low-pass sequencing data, leveraging a high-coverage whole-genome sequencing (WGS) reference panel. Our study included 120 WGS samples (mean coverage 18à—) and 1,500 low-pass sequenced individuals (mean coverage 1.5à—). The individuals with high coverage sequence data were selected as being key ancestors of the current breeding population. The individuals with low pass sequence data were from more recent generations and are thus descendants from the key ancestors. From the WGS panel, we identified over 6 million SNPs and indels and called 49,055 SVs using Manta and 15,613 using smoove. After merging and filtering a high-confidence dataset of 34,274 SVs was left, with micro-chromosomes showing a clear enrichment of SVs relative to macrochromosomes. We first investigated linkage disequilibrium (LD) between SVs and SNPs in the WGS reference panel. Deletions exhibited the highest LD with nearby SNPs, with ~70% of deletions in strong LD (r² >0.8), while duplications and inversions showed lower LD (~40% in strong LD). Interestingly, LD was not strongly dependent on minor allele frequency, indicating that even rare variants can be tagged by SNPs. Before imputation, we calculated genotype likelihoods (GLs) for SNPs using bcftools and for SVs (DEL, DUP, INV) using SVTyper. These GLs were merged and used as input for GLIMPSE2 imputation. To validate the approach, five WGS samples were down sampled to 1.5à— and imputed using the same workflow. Concordance rates between imputed and original genotypes were high across the following three variant types: SNPs (98.4-98.85%), indels (95.88-96.75%), deletions (92.41-93.19%), but somewhat lower for inversions (86.09-89.55%), and duplications (80.53-82.71%). Our results demonstrate that low-pass sequencing combined with genotype likelihood-based imputation provides a powerful and accurate approach to genotype SVs, achieving particularly high accuracy for SNPs, indels, and deletions, and moderate accuracy for duplications and inversions. The resulting high-density dataset of imputed variants represents a valuable resource for downstream applications, including genome-wide association studies and genomic prediction, enabling more comprehensive incorporation of structural variation into population genetic analyses.

Keywords: 2026

How to Cite:

Derks, M., Pook, T., Bink, M., Bickhart, D. & Bouwman, A., (2026) “Accurate structural variation imputation using low-pass sequencing data in a turkey population”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2283276. doi: https://doi.org/10.31274/wcgalp.23495

Rights: 1