Machine learning methods and predictive ability metrics for genome-wide prediction of complex traits

González-Recio, Óscar; Rosa, Guilherme J. M.; Gianola, Daniel

doi:10.1016/j.livsci.2014.05.036

Cited by 130 publications

(109 citation statements)

References 53 publications

Supporting

Mentioning

105

Contrasting

Order By: Relevance

“…Prediction accuracy measured as the correlation is an estimator of the linear relationship between predictions and responses and does not address the bias of predictions, in contrast to the MSE, as noted by González-Recio et al [5]. The optimal number of animals with self-trained phenotypes based on minimizing MSE, was less than 400, as shown in Fig.…”

Section: Resultsmentioning

confidence: 94%

Semi-supervised learning for genomic prediction of novel traits with small reference populations: an application to residual feed intake in dairy cattle

2016

View full text Add to dashboard Cite

BackgroundGenomic prediction for novel traits, which can be costly and labor-intensive to measure, is often hampered by low accuracy due to the limited size of the reference population. As an option to improve prediction accuracy, we introduced a semi-supervised learning strategy known as the self-training model, and applied this method to genomic prediction of residual feed intake (RFI) in dairy cattle.MethodsWe describe a self-training model that is wrapped around a support vector machine (SVM) algorithm, which enables it to use data from animals with and without measured phenotypes. Initially, a SVM model was trained using data from 792 animals with measured RFI phenotypes. Then, the resulting SVM was used to generate self-trained phenotypes for 3000 animals for which RFI measurements were not available. Finally, the SVM model was re-trained using data from up to 3792 animals, including those with measured and self-trained RFI phenotypes.ResultsIncorporation of additional animals with self-trained phenotypes enhanced the accuracy of genomic predictions compared to that of predictions that were derived from the subset of animals with measured phenotypes. The optimal ratio of animals with self-trained phenotypes to animals with measured phenotypes (2.5, 2.0, and 1.8) and the maximum increase achieved in prediction accuracy measured as the correlation between predicted and actual RFI phenotypes (5.9, 4.1, and 2.4%) decreased as the size of the initial training set (300, 400, and 500 animals with measured phenotypes) increased. The optimal number of animals with self-trained phenotypes may be smaller when prediction accuracy is measured as the mean squared error rather than the correlation between predicted and actual RFI phenotypes. ConclusionsOur results demonstrate that semi-supervised learning models that incorporate self-trained phenotypes can achieve genomic prediction accuracies that are comparable to those obtained with models using larger training sets that include only animals with measured phenotypes. Semi-supervised learning can be helpful for genomic prediction of novel traits, such as RFI, for which the size of reference population is limited, in particular, when the animals to be predicted and the animals in the reference population originate from the same herd-environment.

show abstract

Section: Resultsmentioning

confidence: 94%

Semi-supervised learning for genomic prediction of novel traits with small reference populations: an application to residual feed intake in dairy cattle

2016

View full text Add to dashboard Cite

show abstract

“…O'Hara and Sillanpää (2009) suggested that m should be no more than 10-15 times greater than n when using any penalized regression analyses such as those suggested in this review whereas González-Recio et al (2014) has suggested a corresponding m:n ratio cap of 50-100. With a 777K SNP chip as becoming increasingly common for Holstein cattle (Wiggans et al 2011), this conservatively would imply that the number of animals should be no less than, say, 50,000.…”

Section: Increasing Marker Densitiesmentioning

confidence: 90%

“…Various nonparametric approaches to WGP have been more extensively advocated such as kernel regression and various other machine learning approaches (González-Recio et al 2014). Although they are not nearly used as widely as the Bayesian alphabet models described thus far, they have been periodically demonstrated to be superior with respect to predictive accuracy, particularly when epistasis is extensive (Howard et al 2014).…”

Section: Nonparametric Versus Parametric Approaches?mentioning

confidence: 99%

“…Given these concerns, it may be useful to explore other metrics for performance, including those discussed near the end of González-Recio et al (2014) or the Hellinger distance as proposed by Lehermeier et al (2013). Plasmode-based permutations of the data are also useful to address false positive performance rates between competing methods for GWA (Gauldron-Duarte et al 2014).…”

Section: Criteria For Assessing Goodness Of Fit or Superiority Of Permentioning

confidence: 99%

“…This ever expanding library of Bayesian parametric approaches to WGP, with labels that typically involve various letter/symbol suffixes to "Bayes," has been satirically referred to as "Bayesian alphabet" analyses (Gianola 2013;Gianola et al 2009). Criticisms of these approaches have been followed by a stronger advocacy for alternative approaches based on machine learning and/or nonparametric techniques (de Los Campos et al 2009a;De los Campos et al 2010b;González-Recio et al 2014; with the basic premise being that these latter methods are better designed for prediction rather than for inference.…”

Section: An Introduction To Whole Genome Prediction Modelingmentioning

confidence: 99%

See 2 more Smart Citations

Statistical and Computational Challenges in Whole Genome Prediction and Genome-Wide Association Analyses for Plant and Animal Breeding

Tempelman

2015

JABES

View full text Add to dashboard Cite

Whole genome prediction (WGP) modeling and genome-wide association (GWA) analyses are big data issues in agricultural quantitative genetics. Both areas require meaningful input from the statistical scholarly community in order to further improve the accuracy of prediction of genetic merit and inference on putative causal variants as well as improving the computational efficiency of existing methods and algorithms. These concerns have become increasingly critical as new sequencing technologies will only exacerbate current model dimensionality problems. We focus primarily on mixed model and hierarchical Bayesian analyses which have been most commonly pursued by animal and plant breeders for WGP thus far. We draw attention to our observation that many such previous analyses have not carefully inferred upon hyperparameters defined at the top levels of the Bayesian model hierarchy, but simply arbitrarily specify their values. We also reassess previous discussions on WGP model dimensionality, believing that useful data augmentation schemes utilized in various Markov Chain Monte Carlo (MCMC) schemes have led to a general misunderstanding that heavy-tailed or variable selection-based WGP models may be highly parameterized relative to more standard mixed model representations. Computational efficiency is addressed with respect to MCMC and competitive, albeit approximate, alternatives. Furthermore, GWA analyses are reassessed, encouraging a greater reliance on shrinkage-based inferences based on critically chosen priors, instead of potentially nonreproducible fixed effects P valuebased inference.

show abstract

Ideas in Genomic Selection with the Potential to Transform Plant Molecular Breeding

McGowan

Wang

Jia

et al. 2021

Plant Breeding Reviews

View full text Add to dashboard Cite

Machine learning methods and predictive ability metrics for genome-wide prediction of complex traits

Cited by 130 publications

References 53 publications

Semi-supervised learning for genomic prediction of novel traits with small reference populations: an application to residual feed intake in dairy cattle

Semi-supervised learning for genomic prediction of novel traits with small reference populations: an application to residual feed intake in dairy cattle

Statistical and Computational Challenges in Whole Genome Prediction and Genome-Wide Association Analyses for Plant and Animal Breeding

Ideas in Genomic Selection with the Potential to Transform Plant Molecular Breeding

Contact Info

Product

Resources

About