Multi-omics analysis

Can microbiome improve genetic evaluations? Comparison between GBLUP and GO-BLUP in commercial beef populations

---

Abstract

Genetic improvement in cattle depends on the accurate prediction of estimated breeding values (EBVs) for economically important traits. While genomic selection has markedly increased EBV accuracy in recent decades, new sources of biological information may further enhance prediction. The rumen microbiome has emerged as a potential intermediate phenotype linking the host genotype and phenotype. Integrating this metagenomic information into genomic evaluations may therefore improve selection accuracy.The objective of this study is to compare the predictive ability (PA) of the genomic best linear unbiased prediction (GBLUP) model with the genomic-omic BLUP (GO-BLUP) model for carcass and feed efficiency traits in commercial beef cattle. Data included 1,458 British Blue à— Holstein crossbred animals genotyped for 65,782 SNPs, aged 213 ± 33 days at the start of the trial, all fed the same commercial diet and evaluated at a commercial test station. Feed intake and body weight were recorded during a 4-week trial, allowing the estimation of feed conversion ratio (FCR), average daily gain (ADG), daily feed intake (DFI) and residual feed intake (RFI). At the end of the trial, intramuscular fat (IMF), eye muscle area (EMA), and subcutaneous fat thickness (SFT) at the 12th-13th rib was measured by ultrasound. Rumen samples were collected via stomach tube for whole-metagenome sequencing, and carcass weight (CW) was recorded at slaughter.The model was trained using 80% of the animals, with the remaining 20% used as an external testing set with masked phenotypes across 100 random data splits. All performance trait models included sex and batch as fixed effects, and diet composition variables and age as covariates. First, a GBLUP model with a genomic random effect was fitted. Subsequently, a two-step GO-BLUP approach was implemented. In the first step, a GBLUP model including genomic and microbial random effects was fitted to estimate direct genomic and microbiome variances using the GRM and MRM matrices. Estimated microbial values were then used as a phenotype in a second model including sequencing depth as a fixed effect and a genomic random effect to estimate microbiome-mediated genomic variance, thereby partitioning genetic and environmental components. Empirical PA was calculated as the correlation between adjusted phenotypes and EBVs in the test set.Across traits, GO-BLUP yielded higher PA than GBLUP in animals with masked phenotypes for ADG, with a mean improvement of +20% (p UBA2810 and Dialister. No relevant differences were observed for EMA, CW, SFT, IMF, or RFI. These results suggest that integrating rumen metagenomic data into genomic evaluations can improve EBV accuracy for certain commercially relevant traits.

Keywords: 2026

How to Cite:

Saez-Torillo, S., Nguyen, T., Lima, J., Cleveland, M., Roehe, R. & Martínez-Álvaro, M., (2026) “Can microbiome improve genetic evaluations? Comparison between GBLUP and GO-BLUP in commercial beef populations”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2285803. doi: https://doi.org/10.31274/wcgalp.23797

Rights: 1