Skip to main content
. 2022 Oct 31;13:1021182. doi: 10.3389/fpls.2022.1021182

Table 3.

Genomic prediction accuracies per type, per model, per crop class and per trait.

Type Model Red Rot resistance CCS TCH
Random Mixed P 1R P 1R
A BayesA 0.25 0.20 0.19 0.11 0.16
A BayesB 0.26 0.19 0.18 0.11 0.16
A BL 0.25 0.18 0.18 0.11 0.16
A GBLUP 0.24 0.22 0.22 0.11 0.16
A RKHS 0.25 0.22 0.21 0.12 0.16
A RFR 0.18 0.18 0.17 0.11 0.14
M BayesA 0.40 0.15 0.27 0.27 0.20 0.22
M BayesB 0.44 0.48 0.18 0.20 0.17 0.19
M BL 0.41 0.50 0.28 0.28 0.20 0.22
M GBLUP 0.41 0.50 0.28 0.28 0.20 0.22
M RKHS 0.41 0.51 0.29 0.29 0.19 0.22
M RFR 0.46 0.49 0.28 0.28 0.31 0.33
AM BayesA 0.40 0.27 0.28 0.20 0.23
AM BayesB 0.45 0.23 0.23 0.17 0.19
AM BL 0.42 0.28 0.29 0.20 0.22
AM GBLUP 0.40 0.28 0.28 0.18 0.22
AM RKHS 0.40 0.28 0.28 0.17 0.21
AM RFR 0.46 0.27 0.28 0.30 0.32

CCS, commercial cane sugar; TCH, tonnes cane per hectare; RR, red rot. The types of model refer to models (Bayes A, Bayes B, Bayesian Lasso, Ridge regression (GBLUP), Kernal Hilbert spaces (RKHS), Random Forest Regression (RFR), with pedigree data only (A), marker data only (M) and pedigree and marker data combined (AM). For Red Rot resistance, accuracies are given for models assuming random marker effects only (random) and models assuming a mixed model (Mixed), with 29 markers given in Table 3 designated as fixed effects and the remaining markers as random effects.