Table 2  Classification and abbreviations of the models included in Figure 3, A and B
Name (abbreviation)BayesianPenalizedNonparametric
Least-squares regression (LSR)
Bayesian ridge regression (BRR) or RR-BLUPXX
BLUP using a genomic relationship matrix (G-BLUP)XX
Trait-specific BLUP (TA-BLUP)XX
BayesAX
BayesBX
BayesCX
Bayes SSVSX
Bayesian LASSO (BL)X
Double hierarchical generalized linear models (DHGLM)
Least absolute shrinkage and selection operator (LASSO)X
Partial least-squares regression (PLS)X
Principal component regression (PCR)X
Elastic net (EN)X
Reproducing kernel Hilbert spaces regressions (RKHS)XXX
Support vector regression (SVR)XX
BoostingaNANANA
Random forests (RF)X
Neural networks (NN)bXXX
  • The following are early references of the use of the above methods for genomic prediction (references with the original description of some of the methods are also given in earlier sections of this article and in the references given here). LSR, BRR, BayesA, and BayesB, Meuwissen et al. (2001); G-BLUP, VanRaden (2008); TA-BLUP, Zhang et al. (2010); BayesC, Habier et al. (2011); Bayes SSVS, Calus et al. (2008); BL, de los Campos et al. (2009); DHGLM, Shen et al. (2011); LASSO, Usai et al. (2009); PLS and SVR, Moser et al. (2009); PCR, Solberg et al. (2009); EN, croiseau et al. (2011); RKHS, gianola et al. (2006); Boosting, González-Recio et al. (2010); RF, González-Recio and Forni (2011); and NN, Okut et al. (2011).

  • a Boosting as an estimation technique could be applied to any method, Bayesian or penalized, parametric or nonparametric.

  • b NN could be implemented in a nonpenalized, penalized, or Bayesian framework.