Incorporating transcriptomic data into genomic prediction models to improve the prediction accuracy of phenotypes of...

Since genomic selection has been established in animal breeding, attention has turned towards other omics layers that are seen as promising to improve prediction accuracy. Transcriptomic data provide insights into gene expression patterns, which are shaped by both genetic and environmental factors, offering a more comprehensive understanding of the expression of phenotypes. This study utilized various statistical methods to assess the applicability of transcriptomic data derived from intestinal tissue to the prediction of efficiency-related phenotypes. The focus was on formal derivation of the previously described GTCBLUP model, which was adapted to create GTCBLUPi and compared with other BLUP models. The GTCBLUPi model addresses redundant information between genomic and transcriptomic information. We compared estimated variance components and accuracies of prediction of phenotypes for efficiency-related traits in an F2 cross of 480 Japanese quail using different models. Additionally, we estimated transcriptomic correlations between the traits using animal effects based on transcriptomic similarity, and the effects of individual transcript abundances on the phenotypes.

This study showed that transcript abundances from the ileum explain a larger portion of the phenotypic variance of the traits than host genetics. Models incorporating both genetic and transcriptomic information outperformed those using only one type of information, with regard to the phenotypic variances explained. The combination of both data types resulted in higher trait prediction accuracies, confirming that transcriptomic information complements genetic data effectively. The derived GTCBLUPi model proved to be a suitable framework for integrating both information types. Additionally, polygenic backgrounds were identified for the traits studied based on transcriptomic profiles, along with high transcriptomic correlations between the traits.

Transcriptomic data account for a high portion of phenotypic expression for all phenotypes and incorporating them enables more accurate predictions of phenotypes for efficiency and performance traits. Models that integrate both genetic and transcriptomic information are the most effective, offering valuable insights for improving phenotype prediction accuracy and insights in biological mechanisms underlying phenotypic variation of traits.