Figure 1.
Histogram of the offdiagonal elements of the genomic relationship matrix .
The genomic relationship matrix was calculated according to [8] using 157 lines and 2.5 million SNPs.
Figure 2.
Heatmap of the genomic relationship matrix .
The genomic relationship matrix was calculated according to [8] using 157 lines and 2.5 million SNPs. The “S” after the line-ID indicates that the line belongs to the set of lines for which phenotypic records for startle response were also available (in addition to the phenotypic records of starvation resistance).
Figure 3.
Accuracy of prediction of GBLUP for CVs with different numbers of lines in the training set.
Each boxplot illustrates the average accuracies for 20 replicates of the CV procedure using GBLUP. The left (right) plot shows accuracies for starvation resistance (startle response). The solid line is the curve of [20] fitted to the empirical data, which results in estimates of and
for starvation resistance and startle response. All 2.5 million SNPs were used to construct the genomic relationship matrix in the GBLUP model.
Table 1.
Average correlations between predicted genetic values and observed phenotypes for different CV procedures with GBLUP and different traits.
Figure 4.
Predictive ability of 5-fold CV with GBLUP for starvation resistance using different numbers of SNPs.
Each boxplot shows the average predictive abilities for 20 replicates of 5-fold CV using GBLUP. For the CVs leading to the -th boxplot, every 2
-th SNP was used to build the genomic relationship matrix
according to [8]. This was done for the thinning factors
0
10. The red dots indicate the average predictive abilities.
Figure 5.
The distribution of between SNP neighbors for different SNP densities.
For the -th stacked bar, every
-th SNP was used,
0
10. Then, the distribution of
for the resulting SNP neighbors was calculated.
Figure 6.
Predictive ability for GBLUP versus BayesB using phenotypic values of starvation resistance.
Predictive abilities are plotted for 20 replicates of a 5-fold CV, each replicate consisting of 5 corresponding folds of CV.
Figure 7.
Distribution of absolute SNP effects.
The density of the sum of the absolute SNP effects from GBLUP is plotted for sliding windows of 100 adjacent SNPs covering the whole genome (black) and for windows around the 75 most significant SNPs (red) according to the GWAS of [27]. The left (right) plot shows the densities for starvation resistance (startle response). The blue vertical line indicates the 90% quantile of the black density function.