Answers are grounded in this paper and cite their sources.
Genomic selection (GS) in beef cattle is constrained by small reference populations and low biological interpretability of conventional models. This study developed three mixed-model approaches: GBLUPmix (GBLUP with a mixed reference population), ssGBLUPmix (single-step GBLUP with a mixed reference population), and wGBLUPmix (weighted GBLUP with a mixed reference population) to test whether a functionally informed genomic relationship matrix (Gfunc) built from Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) annotations improves cross-population prediction accuracy. Three simulated populations (PopA, PopB, PopC) were used to construct pooled reference populations. We randomly sampled 1000 to 5000 individuals from each source population. Equal numbers of individuals were sampled from PopA, PopB and PopC and combined, producing pooled reference populations of 3000, 6000, 9000, 12,000 and 15,000 individuals in total. Functional SNPs based on GO and KEGG annotations were used to construct the Gfunc matrix and assess model prediction performance. Results showed that wGBLUPmix with KEGG-informed Gfunc consistently achieved the highest prediction accuracy and robustness across population sizes and genetic backgrounds. KEGG annotations outperformed GO annotations, especially in larger and more genetically divergent reference sets. To further verify the practical reliability of the proposed framework, we conducted validation on real beef cattle populations, including imported Angus, Limousin, and indigenous Sanhe cattle with weaning weight (WWT) as the target trait. Real-population validation confirmed that pathway-informed functional matrices effectively mitigate prediction loss caused by inter-breed genetic differentiation. These findings demonstrate that integrating KEGG pathway annotations into weighted genomic prediction models provides a practical strategy to improve multi-breed GS for small local cattle breeds. Future work will combine Gfunc with hierarchical Bayesian models to separate shared and breed-specific SNP effects in complex breeding systems.
No takes yet. Share an insight, caveat, or question.
Zhou et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: