@article{wang_maltecca_tiezzi_huang_jiang_2023, title={123 Benchmarking of Artificial Neural Network Models for Genomic Prediction of Quantitative Traits in Pigs}, volume={101}, ISSN={0021-8812 1525-3163}, url={http://dx.doi.org/10.1093/jas/skad281.022}, DOI={10.1093/jas/skad281.022}, abstractNote={Abstract Artificial neural networks (ANN) are a type of machine learning model that has been applied to various genomic problems, with the ability to learn non-linear relationships and model high-dimensional data. ANNs also have the potential in genomic prediction by capturing the intricate relationship between genetic variants and phenotypes. However, there is currently a limited effort to investigate the performance and feasibility of ANNs for pig genomic predictions. In this study, we evaluated the predictive performance of TensorFlow’s ANN models with one-layer, two-layer, and three-layer structures (with zero, one, and two hidden layers, respectively), in comparison with five linear methods, including GBLUP, LDAK, BayesR, SLEMM and scikit-learn’s ridge regression using data of six quantitative traits including off-test body weight (WT), off-test back fat thickness (BF), off-test loin muscle depth (MS), number of piglets born alive (NBA), number of piglets born dead (NBD), and number of piglets weaned (NW). Furthermore, we assessed the computational efficiency of ANNs on both CPU and GPU. The benchmarking was based on cross-validations of 26,190 genotyped pigs. We employed hyperband tuning to optimize the hyper-parameters and select the best model among one-layer, two-layer, and three-layer structures. Results showed that the one-layer structure, which is equivalent to ridge regression, yielded the best performance comparable to that of GBLUP. Using the optimal hyper-parameters for two-layer and three-layer structures, ANNs underperformed GBLUP in terms of accuracy. Of the five linear methods, BayesR and SLEMM performed similarly and the best, followed by LDAK, scikit-learn’s ridge regression, and GBLUP. Moreover, SLEMM was the fastest, which completed training with 21k individuals and 30k SNPs in 2.6 minutes. Compared with CPUs, GPUs exhibited a comparable computational speed for one-layer ANN but offered significant gains in computational efficiency for multi-layer ANNs. Based on our analysis of optimal hyper-parameters for two-layer ANN with BF, we found that using a GPU can lead to a five-fold increase in processing speed compared with using a conventional CPU, but it is still slower than GBLUP. In addition, hyper-parameter tuning (particularly for L2 regularization and the number of dense units in hidden layers) is critical for improving the genomic prediction accuracy in pigs. In conclusion, we found ANN with up to three layers could not improve genomic predictions compared with routine linear methods for pig quantitative traits.}, number={Supplement_3}, journal={Journal of Animal Science}, publisher={Oxford University Press (OUP)}, author={Wang, Junjian and Maltecca, Christian and Tiezzi, Francesco and Huang, Yijian and Jiang, Jicai}, year={2023}, month={Nov}, pages={17–18} }
 @article{marceau_wang_iqbal_jiang_liu_ma_2023, title={Investigation of lncRNA in <i>Bos taurus</i> Mammary Tissue during Dry and Lactation Periods}, volume={14}, ISSN={["2073-4425"]}, url={https://doi.org/10.3390/genes14091789}, DOI={10.3390/genes14091789}, abstractNote={This study aims to collect RNA-Seq data from Bos taurus samples representing dry and lactating mammary tissue, identify lncRNA transcripts, and analyze findings for their features and functional annotation. This allows for connections to be drawn between lncRNA and the lactation process. RNA-Seq data from 103 samples of Bos taurus mammary tissue were gathered from publicly available databases (60 dry, 43 lactating). The samples were filtered to reveal 214 dry mammary lncRNA transcripts and 517 lactating mammary lncRNA transcripts. The lncRNAs met common lncRNA characteristics such as shorter length, fewer exons, lower expression levels, and less sequence conservation when compared to the genome. Interestingly, several lncRNAs showed sequence similarity to genes associated with strong hair keratin intermediate filaments. Human breast cancer research has associated strong hair keratin filaments with mammary tissue cellular resilience. The lncRNAs were also associated with several genes/proteins that linked to pregnancy using expression correlation and gene ontology. Such findings indicate that there are crucial relationships between the lncRNAs found in mammary tissue and the development of the tissue, to meet both the animal’s needs and our own production needs; these relationships should be further investigated to ensure that we continue to breed the most resilient, efficient dairy cattle.}, number={9}, journal={GENES}, author={Marceau, Alexis and Wang, Junjian and Iqbal, Victoria and Jiang, Jicai and Liu, George E. and Ma, Li}, year={2023}, month={Sep} }
 @article{wang_hicks_jiang_liu_2022, title={Transcriptome Analysis of Chicken Reveals the Impact of Herpesvirus of Turkeys on Bursa RNA Expression in Marek's Disease Virus Infections}, volume={100}, ISSN={["1525-3163"]}, DOI={10.1093/jas/skac247.392}, abstractNote={Abstract Marek’s disease virus (MDV) is an oncogenic herpesvirus that causes various clinical syndromes in chicken. MDV early infection induces a transient immunosuppression and harbored in B cells of the bursa during the cytolysis phase of its replication cycle. One of the most commonly used commercial vaccines is Herpesvirus of Turkeys (HVT), which is a nonpathogenic virus of domestic turkey and may influence the expression of RNA. The aim of this study is to characterize the regulation of bursa gene in MDV infections and the impact of HVT vaccination on RNA expression in MDV-infected chickens. We used RNA-seq on the bursa samples to compare the transcriptome differences among MDV-infected chickens, HVT-infected chickens, co-infected chickens and uninfected control groups at 4, 7, 14 and 21 days post infection. Meanwhile, we also compared the expression at three different time points to examine alterations in the expression pattern. The result of differential gene expression showed that 14 days post infection might be the point in time when HVT worked. At false discovery (FDR) < 0.05 and fold change (FC) ≥2, we found 745, 218 and 76 genes in MDV-infected chickens, HVT-infected chickens and co-infected chickens respectively as differentially expressed compared with control group and 713 genes between MDV-infected chickens and co-infected chickens at 14 dpi. KEGG and GO enrichment analysis showed that these genes were highly enriched for Lysosome, immune response, inflammatory response, plasma membrane and so on. Overall, these findings help to better understand host-pathogen interaction in the bursa and elucidate the mechanism how HVT contribute to resist MDV. Further investigation of the roles of these candidate genes and signaling pathways in the regulation of MDV-HVT interaction may lead new directions for the development of drugs or cultivation of highly MDV resistant chickens.}, journal={JOURNAL OF ANIMAL SCIENCE}, author={Wang, Junjian and Hicks, Julie and Jiang, Jicai and Liu, Hsiao-Ching}, year={2022}, month={Oct}, pages={215–215} }