Genetic gain & Inbreeding

Predicting Average Progeny Inbreeding in Canadian Holstein Cattle Using Expected Runs of Homozygosity

---

Abstract

Managing genetic diversity is essential for sustaining long-term genetic progress and to mitigate the adverse effects of inbreeding in livestock. Runs of homozygosity (ROH) have been shown to outperform both pedigree-based and SNP-based approaches for quantifying inbreeding. The objectives of this study were to: 1) predict and validate average progeny inbreeding using expected ROH derived from parental genotypes in Canadian Holstein dairy cattlle, and 2) evaluate the influence of haplotype phasing on expected ROH prediction accuracy. Genotype data included 658,219 Canadian Holstein dairy cattle genotyped using various SNP panels. Pedigree records were traced back to 1762, encompassing 1,605,697 individuals. Genotypes were imputed to the Illumina Bovine SNP50 V2 panel using FImpute. Animals with both parents and all four grand-parents genotyped were selected, resulting in a final dataset of 26,836 progeny from 1,120 sires and 18,645 dams. SNPs with minor allele frequency (MAF) less than 0.01 or located on sex chromosome were excluded, leaving 42,693 SNPs for analysis. Parental genotypes were phased using FImpute to reconstruct haplotypes. For each mating pair, four pseudo-genotypes were generated by systematically combining the two haplotypes from each parent. ROH were identified in both pseudo-genotypes and real progeny using GARLIC software, employing a sliding window of 20 SNPs. ROH were classified into short, medium, and long classes via Gaussian mixture modeling, and ROH-based inbreeding was calculated as the proportion of the genome within ROH. Expected ROH-based inbreeding for each mating pair was computed as the average of the four pseudo-genotype estimates. ROH length distributions and inbreeding estimates from pseudo-genotypes were compared to those observed in real progeny to evaluate prediction accuracy. To assess the influence of haplotype phasing on expected ROH prediction, parental haplotypes were phased under three additional scenarios. In scenario S1, all progeny of the candidates were removed from the reference population. In scenario S2, both progeny and dams of the candidates were removed. In scenario S3, both parents and progeny of the candidates were removed. The distribution of ROH lengths identified in pseudo-genotype closely mirrored that observed in real progeny. Expected ROH-based inbreeding showed strong agreement with realized ROH-based inbreeding across all ROH length classes. Correlations increased with ROH length with r = 0.62 for short ROH, r = 0.73 for medium ROH, r = 0.76 for long ROH, and r = 0.83 for total ROH. Regression slopes were close to unity (0.96-0.99), and intercepts were near zero, indicating minimal bias. Haplotype phasing accuracy exceeded 99.35% across different scenarios, resulting in negligible impact on expected ROH prediction. Findings from this study suggest that expected ROH can be effectively used for pre-mating decisions to manage future inbreeding in Canadian Holstein breeding programs.

Keywords: 2026

How to Cite:

Chen, L., Sullivan, P., Fleming, A., Van Doormaal, B., Jaton, C. & Miglior, F., (2026) “Predicting Average Progeny Inbreeding in Canadian Holstein Cattle Using Expected Runs of Homozygosity”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2272435. doi: https://doi.org/10.31274/wcgalp.23413

Rights: 1