Breeding goals & design

Evaluation of the genotype imputation accuracy between SNP Chips in different densities and their impacts on (g)EBV prediction in pig breeding

---

Abstract

Genotype imputation accuracy and its impacts on estimated breeding values (EBVs) were evaluated in different data sets, with aims to optimize the cost-effective genotyping strategy and maximize the genetic gain for a pig breeding company. A total of 6,626 purebred Yorkshire were genotyped by 50K SNP Chip. A subset of 124 was also genotyped by 5K and 75K SNP Chips. When imputing 5K to 50K, the overall concordance rate (CR) per-individual was 98.72%, with the highest accuracy observed for individuals with at least one parent represented in the reference set. The overall dosage squared correlation (R²) was 0.976, with the highest accuracy for SNPs with minor allele frequency (MAF) ≥ 0.2 and the lowest for SNPs with MAF < 0.05. In contrast, imputing 50K to 75K using 10-fold cross-validation resulted in lower accuracy, with an average CR per-individual of 84.94% and an overall R² of 0.475. The (g)EBVs for age (AGE) and loin muscle depth (LMD) at 120kg were evaluated through a bi-variate animal model. Compared with pedigree-based EBVs, incorporating real 50K genotypes (gEBV1) increased accuracy from 0.39 to 0.46 for AGE and from 0.26 to 0.69 for LMD. Using imputed 50K from 5K (gEBV2) or imputed 75K from 50K (gEBV3) resulted in a slight decline in prediction accuracy compared with gEBV1. And the dispersion bias decreased further from 1 for both traits, indicating the inflation of the (g)EBVs when using imputed genotypes. In practical breeding program, with an existing of a large and well-connected 50K reference population, a cost-effective strategy would be using the 5K SNP chip for large-scale genotyping, then imputing to 50K for genomic prediction. Continually genotyping a small number of replacement parents using 50K Chip can further improve the imputation performance, consequently to accelerate the genetic gain for a commercial pig breeding company.

Keywords: 2026

How to Cite:

Zhang, C., Jiang, W. & Wu, R., (2026) “Evaluation of the genotype imputation accuracy between SNP Chips in different densities and their impacts on (g)EBV prediction in pig breeding”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2296026. doi: https://doi.org/10.31274/wcgalp.24388

Rights: 1