Skip to main content
Fig. 2 | Genome Biology

Fig. 2

From: LightGBM: accelerated genomically designed crop breeding through ensemble learning

Fig. 2

Evaluations of GB and rrBLUP precision on different predictive frameworks. a Schematic illustration of the four ways partitioning the training and testing samples. M and P: training samples cover the maternal and paternal genotypes of testing samples. M only: training samples only cover the maternal genotypes of testing samples. P only: training samples only cover the paternal genotypes of testing samples. Neither: training samples cover neither of the maternal and paternal genotypes of the testing samples. b Evaluations of GB and rrBLUP precision on the four different predictive frameworks. c Evaluations of GB and rrBLUP precision between two fully genotyped F1 populations. Left three columns: 1428 Zheng58 F1s as training samples to predict Jing724 F1s; Right three columns: 1428 Jing724 F1s as training samples to predict Zheng58 F1s

Back to article page