Table 2 Genomic selection accuracy (r), standard deviation of accuracy (σ), relative efficiency per cycle (REc), and relative efficiency per year (REy) of three genomic selection models using a TP of 470 wheat lines and 10-fold cross-validation for four traits: grain yield over all six environments (GYA), grain yield at Northwest Ohio environments (4 environments, GYN), grain yield at Wooster, Ohio (2 environments, GYW), flour yield (FY), softness equivalence (SE), and Fusarium Head Blight index (FHBI)
Ridge-Regression BLUPaRandom ForestBayesian LASSO
TraitrbσREccREyrdrσREcREyrrσREcREyr
GYA0.450.010.584.10.480.010.624.30.120.020.151.1
GYN0.410.010.584.10.420.010.594.10.140.020.201.4
GYW0.570.010.674.70.570.010.694.80.570.010.684.8
FY0.620.010.673.40.630.010.683.40.220.010.241.2
SE0.510.010.532.70.490.010.522.60.060.020.060.3
FHB0.350.010.462.30.370.010.482.40.170.020.221.1
  • a RRBLUP and Bayesian LASSO (BLR) were run for 1500 cycles and RF was run for 500 cycles.

  • b Accuracy is the Pearson’s correlation between the phenotype and the genomic estimated breeding value.

  • c Relative efficiency per cycle calculated by Embedded Image.

  • d Relative efficiency per year calculated as REc times the ratio of years in a cycle of phenotypic selection to years in a cycle of genomic selection.