Genetically regulated omics integration improves genomic prediction of complex traits in pigs

Area of Science:

• Genetics
• Genomics
• Bioinformatics

Background:

• Complex traits are influenced by DNA variation, but intermediate omics data (e.g., transcriptomes) are noisy and mix genetic/non-genetic signals.
• Accurate prediction of complex traits requires disentangling genetic architecture, including additive and dominance effects.

Purpose of the Study:

• To develop and evaluate a novel framework, the Genetically Regulated Additive and Dominance (GRAD) model, for predicting complex traits.
• To improve prediction accuracy by explicitly modeling genetically regulated additive and dominance effects from omics data.

Main Methods:

• Developed the GRAD model, a single-step framework that decomposes omics features into genetically regulated additive and dominance components.
• Jointly modeled additive and dominance effects at genomic and intermediate-omics levels.
• Validated GRAD using simulations and a Landrace pig dataset.

Main Results:

• GRAD achieved up to 21.98% average accuracy gain in simulations compared to standard genomic prediction.
• GRAD outperformed models using raw omics or only additive omics signals.
• In pigs, GRAD improved prediction accuracy for growth and reproductive traits by 7.4% and 14.9%.

Conclusions:

• Explicitly extracting genetically regulated additive and dominance signals from omics data enhances prediction accuracy.
• The practical utility of GRAD depends on the proportion of profiled animals and the relevance of sampled tissues to the target trait.