Extend mixed models to multilayer neural networks for genomic prediction including intermediate omics data

Zhao, Tianjing; Zeng, Jian; Cheng, Hao

doi:10.1093/genetics/iyac034

Cited by 18 publications

(13 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This integration can then be applied to meta- and colocalization analyses (46,47), facilitating a more comprehensive analysis of gene regulatory mechanisms. Multi-omics integration will also contribute to the development of accurate artificial intelligence algorithms (48,49).…”

Section: Discussionmentioning

confidence: 99%

Integrated Global Chicken Reference Panel From 13,187 Chicken Genomes

Zhu,

Wang,

et al. 2023

Preprint

View full text Add to dashboard Cite

Chickens are a crucial source of protein for humans and a popular model animal for bird research. Despite the emergence of imputation as a reliable genotyping strategy for large populations, the lack of a high-quality chicken reference panel has hindered progress in chicken genome research. To address this issue, here we introduce the first phase of the 100 K Global Chicken Reference Panel Project (100 K GCRPP). The project includes 13,187 samples and provides services for varied applications on its website (http://farmrefpanel.com/GCRP/). Currently, two panels are available: a Comprehensive Mix Panel (CMP) for domestication diversity research and a Commercial Breed Panel (CBP) for breeding broilers specifically. Evaluation of genotype imputation quality showed that CMP had the highest imputation accuracy compared to imputation using existing chicken panel in animal SNPAtlas, whereas CBP performed stably in the imputation of commercial populations. Additionally, we found that genome-wide association studies using GCRP-imputed data, whether on simulated or real phenotypes, exhibited greater statistical power. In conclusion, our study indicates that the GCRP effectively fills the gap in high-quality reference panels for chickens, providing an effective imputation platform for future genetic and breeding research.

show abstract

Section: Discussionmentioning

confidence: 99%

Integrated Global Chicken Reference Panel From 13,187 Chicken Genomes

Zhu,

Wang,

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…However, based on the observed decrease in costs of genotyping which has enabled large-scale genotyping, we may expect similar developments for the costs of transcriptomics and other intermediate phenotypes in the near future ( Uzbas et al 2019 ). At the same time, there have been some recent model developments that enable including other omics data in genomic prediction, when these other omics data are not available for all animals ( Christensen et al 2021 ; Zhao et al 2022 ).…”

Section: Discussionmentioning

confidence: 99%

Adding gene transcripts into genomic prediction improves accuracy and reveals sampling time dependence

Perez

Bink

Svenson

et al. 2022

G3 Genes|Genomes|Genetics

View full text Add to dashboard Cite

Recent developments allowed generating multiple high quality ‘omics’ data that could increase predictive performance of genomic prediction for phenotypes and genetic merit in animals and plants. Here we have assessed the performance of parametric and non-parametric models that leverage transcriptomics in genomic prediction for 13 complex traits recorded in 478 animals from an outbred mouse population. Parametric models were implemented using best linear unbiased prediction (BLUP), while non-parametric models were implemented using the gradient boosting machine algorithm (GBM). We also propose a new model named GTCBLUP that aims to remove between-omics-layer covariance from predictors, whereas its counterpart GTBLUP does not do that. While GBM models captured more phenotypic variation, their predictive performance did not exceed the BLUP models for most traits. Models leveraging gene transcripts captured higher proportions of the phenotypic variance for almost all traits when these were measured closer to the moment of measuring gene transcripts in the liver. In most cases, the combination of layers was not able to outperform the best single-omics models to predict phenotypes. Using only gene transcripts, the GBM model was able to outperform BLUP for most traits except body weight, but the same pattern was not observed when using both SNP genotypes and gene transcripts. Although the GTCBLUP model was not able to produce the most accurate phenotypic predictions, it showed highest accuracies for breeding values for 9 out of 13 traits. We recommend using the GTBLUP model for prediction of phenotypes and using the GTCBLUP for prediction of breeding values.

show abstract

“…Indeed, different environmental factors in nature likely act in concert in inducing epigenetic alterations, and the investigation of these interactions is of fundamental importance to better understand the complexity of the process. In this context, artificial neural networks (ANNs) are a solid method to overcome the intrinsic limitations of standard statistical methodologies to investigate non-linear associations that characterize complex biological systems, and are successfully applied in genomics and epigenomics research [ 28 , 29 , 30 ]. In particular, the auto-contractive map (Auto-CM) is a powerful data mining system able to define the strength of the association of each variable with all the others and to visually show the map of their main connections.…”

Section: Introductionmentioning

confidence: 99%

Prenatal Environmental Stressors and DNA Methylation Levels in Placenta and Peripheral Tissues of Mothers and Neonates Evaluated by Applying Artificial Neural Networks

Stoccoro

Nicolì

Coppedè

et al. 2023

Genes

View full text Add to dashboard Cite

Exposure to environmental stressors during pregnancy plays an important role in influencing subsequent susceptibility to certain chronic diseases through the modulation of epigenetic mechanisms, including DNA methylation. Our aim was to explore the connections between environmental exposures during gestation with DNA methylation of placental cells, maternal and neonatal buccal cells by applying artificial neural networks (ANNs). A total of 28 mother–infant pairs were enrolled. Data on gestational exposure to adverse environmental factors and on mother health status were collected through the administration of a questionnaire. DNA methylation analyses at both gene-specific and global level were analyzed in placentas, maternal and neonatal buccal cells. In the placenta, the concentrations of various metals and dioxins were also analyzed. Analysis of ANNs revealed that suboptimal birth weight is associated with placental H19 methylation, maternal stress during pregnancy with methylation levels of NR3C1 and BDNF in placentas and mother’s buccal DNA, respectively, and exposure to air pollutants with maternal MGMT methylation. Associations were also observed between placental concentrations of lead, chromium, cadmium and mercury with methylation levels of OXTR in placentas, HSD11B2 in maternal buccal cells and placentas, MECP2 in neonatal buccal cells, and MTHFR in maternal buccal cells. Furthermore, dioxin concentrations were associated with placental RELN, neonatal HSD11B2 and maternal H19 gene methylation levels. Current results suggest that exposure of pregnant women to environmental stressors during pregnancy could induce aberrant methylation levels in genes linked to several pathways important for embryogenesis in both the placenta, potentially affecting foetal development, and in the peripheral tissues of mothers and infants, potentially providing peripheral biomarkers of environmental exposure.

show abstract

Extend mixed models to multilayer neural networks for genomic prediction including intermediate omics data

Cited by 18 publications

References 54 publications

Integrated Global Chicken Reference Panel From 13,187 Chicken Genomes

Integrated Global Chicken Reference Panel From 13,187 Chicken Genomes

Adding gene transcripts into genomic prediction improves accuracy and reveals sampling time dependence

Prenatal Environmental Stressors and DNA Methylation Levels in Placenta and Peripheral Tissues of Mothers and Neonates Evaluated by Applying Artificial Neural Networks

Contact Info

Product

Resources

About