Application of deep learning with bivariate models for genomic prediction of sow lifetime productivity-related traits

• DOI: 10.5713/ab.23.0264
• Publication Date: 2024-01-14T14:17:57Z
• Authors: Joon-Ki Hong; Yong-Min Kim; Eun-Seok Cho; Jae-Bong Lee; Young-Sin Kim; Hee-Bok Park
• Keywords: pig; 0303 health sciences; 03 medical and health sciences; QL1-991; epistatic interaction; convolutional neural network; deep learning; sow lifetime productivity; Zoology; genomic prediction; Article

Objective: Pig breeders cannot obtain phenotypic information at the time of selection for sow lifetime productivity (SLP). They would benefit from obtaining genetic candidate sows. Genomic data interpreted using deep learning (DL) techniques could contribute to improvement SLP maximize farm profitability because DL models capture nonlinear effects such as dominance and epistasis more efficiently than conventional genomic prediction methods based on linear models. This study aimed investigate usefulness two SLP-related traits; number litters (LNL) pig production (LPP).Methods: Two bivariate models, convolutional neural network (CNN) local (LCNN), were compared with (i.e., best unbiased prediction, Bayesian ridge regression, Bayes A, B). Phenotype pedigree collected 40,011 sows that had husbandry records. Among these, 3,652 pigs genotyped PorcineSNP60K BeadChip.Results: The predictive correlation LNL was obtained CNN (0.28), followed by LCNN (0.26) (approximately 0.21). For LPP, also (0.29), (0.27) 0.25). A similar trend observed mean squared error traits.Conclusion: provides an example a can outperform against model-based approaches when interaction components are important LPP exhibited strong epistatic components. Additionally, our results suggest applying accuracy utilizing between LPP.