Table 1. Comparison of mean R 2 and mean number of significant SNPs across 100 cross-validation runs, calculated for two models and four traits.
Trait |
R2val |
R2train |
Number of significant SNPs |
|||
---|---|---|---|---|---|---|
'IBS-M' | 'IBD-M × F' | 'IBS-M' | 'IBD-M × F' | 'IBS-M' | 'IBD-M × F' | |
HEA | 0.62 | 0.63 | 0.85 | 0.76 | 80 | 6 |
TGW | 0.52 | 0.31 | 0.77 | 0.40 | 74 | 2 |
THR | 0.78 | 0.82 | 0.89 | 0.86 | 54 | 3 |
GrCol | 0.71 | 0.85 | 0.93 | 0.95 | 80 | 5 |
Mean | 0.66 | 0.65 | 0.86 | 0.74 | 72 | 4 |
Abbreviations: GrCol, grain colour; HEA, heading; IBD, identical by descent; IBS, identical by state; R2train, model fit of the training set; R2val, prediction ability; SNP, single-nucleotide polymorphism; TGW, thousand grain weight; THR, threshability.
Traits studied: HEA, TGW, THR and GrCo; Models applied: only marker main effects based on IBS (‘IBS-M’) and only marker-by-family effects based on IBD genotypes (‘IBD-M × F’). The last row indicates the mean across traits.