Multi-omics analysis

Enhancing prediction accuracy for pig production traits with multi-omic mixed models

---

Abstract

Current genomic prediction of breeding values or phenotypes typically relies on single-nucleotide polymorphism (SNP) data from microarrays, while correcting for fixed effects. Advances in biotechnology now enable the large-scale collection of complementary omics data-such as transcriptomics and epigenomics-at decreasing costs. However, the extent to which these added molecular layers improve prediction accuracy compared to SNP-based models is still an open question. We evaluated the impact of incorporating RNA transcript abundances and DNA methylation levels as random effects in linear mixed models for the prediction of three pig phenotypes: backfat thickness (BFT), average daily gain (ADG), and residual feed intake (RFI). Transcriptomic (RNA-sequencing) and methylomic (reduced representation bisulfite sequencing) data were obtained from loin muscle biopsies of 443 DanBred pigs of three breeds (Duroc, Landrace, and Yorkshire). Using these data and phenotypic records, we fit models including genomic, transcriptomic, and methylomic effects. Prediction accuracy was evaluated using leave-one-breed-out (LOBO) and leave-one-out (LOO) cross-validation schemes. When animals from all breeds were used, the predictive accuracy (measured by the Pearson correlation coefficient between predicted values and corrected phenotypes) for ADG ranged from 0.12 with genomic BLUP (GBLUP) to 0.37 with genomic-transcriptomic-methylomic BLUP (GTMBLUP). For BFT, accuracies ranged from 0.18 with methylomic BLUP (MBLUP) to 0.39 with genomic-transcriptomic BLUP (GTBLUP) and GTMBLUP. For RFI, accuracies ranged from 0.19 (GBLUP, MBLUP) to 0.58 (GTBLUP). Transcriptomic (TBLUP) and genomic-methylomic (GMBLUP) models showed intermediate performance. The LOBO cross-validation scheme produced similar patterns of model performance. Overall, adding transcriptomic and methylomic data improved prediction accuracy relative to GBLUP. TBLUP was the strongest single-omic model, while MBLUP gave lower prediction accuracies. The combined GTMBLUP model, which included all omics, achieved the highest accuracies, although the gains over GTBLUP were modest and not statistically significant.

Keywords: 2026

How to Cite:

Ayres, L., Bovenhuis, H., Vourlaki, I., Piles, M., Luigi-Sierra, M., Karlskov-Mortensen, P. & Calus, M., (2026) “Enhancing prediction accuracy for pig production traits with multi-omic mixed models”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2284294. doi: https://doi.org/10.31274/wcgalp.23561

Rights: 1