A Two-Stage Method for Improving the Prediction Accuracy of Complex Traits by Incorporating Genotype by Environment Interactions in Brassica napus
Table 4
The prediction accuracy (standard deviation, SD) of four methods for OC.
Environment
LASSO
OLS post-LASSO
Stratified GBLUP
M × E GBLUP
E7
0.7552 (0.0060)
0.8665 (0.0041)
0.5956 (0.0075)
0.7784 (0.0057)
N3
0.8239 (0.0043)
0.9097 (0.0030)
0.6721 (0.0065)
0.8514 (0.0033)
N4
0.7594 (0.0049)
0.9153 (0.0027)
0.6261 (0.0078)
0.8048 (0.0045)
N6
0.8255 (0.0038)
0.9130 (0.0025)
0.7300 (0.0055)
0.8729 (0.0030)
N7
0.7625 (0.0073)
0.8866 (0.0044)
0.5837 (0.0096)
0.7859 (0.0065)
S3
0.8127 (0.0037)
0.9175 (0.0021)
0.6767 (0.0063)
0.8336 (0.0034)
S4
0.7809 (0.0048)
0.9114 (0.0028)
0.6600 (0.0068)
0.8372 (0.0042)
S5
0.7189 (0.0065)
0.9030 (0.0029)
0.5806 (0.0075)
0.7990 (0.0040)
S6
0.7174 (0.0045)
0.9037 (0.0028)
0.5625 (0.0070)
0.7846 (0.0035)
S7
0.7986 (0.0047)
0.9188 (0.0026)
0.6741 (0.0059)
0.8572 (0.0034)
Average
0.7755 (0.0051)
0.9045 (0.0030)
0.6361 (0.0070)
0.8205 (0.0042)
OC, oil content; LASSO, least absolute shrinkage and selection operator; OLS, ordinary least squares; GBLUP, genomic best linear unbiased prediction; M × E, marker × environment interaction.
