Table 2.
Performance of e‐net and random forest models for trait prediction
| Trait | Transcripts | Regulators | Metabolites | Both | |
|---|---|---|---|---|---|
| Pooled R2 | |||||
| Blade 16 length | Elastic Net | 0.315 (0.002) | 0.313 (0.002) | 0.359 (0.002) | 0.305 (0.002) |
| Random Forest | 0.314 (0.002) | 0.421 (0.002) | 0.308 (0.002) | 0.323 (0.002) | |
| Blade 16 width | Elastic Net | 0.490 (0.002) | 0.459 (0.002) | 0.648 (0.002) | 0.642 (0.002) |
| Random Forest | 0.334 (0.002) | 0.272 (0.002) | 0.496 (0.002) | 0.419 (0.002) | |
| Husk Leaf length | Elastic Net | 0.458 (0.002) | 0.415 (0.002) | 0.509 (0.002) | 0.476 (0.002) |
| Random Forest | 0.258 (0.002) | 0.340 (0.002) | 0.555 (0.002) | 0.450 (0.002) | |
| Ear length | Elastic Net | 0.235 (0.002) | 0.126 (0.006) | 0.208 (0.002) | 0.279 (0.002) |
| Random Forest | 0.100 (0.006) | 0.065 (0.018) | 0.287 (0.002) | 0.131 (0.004) | |
| Plant height | Elastic Net | 0.058 (0.010) | 0.026 (0.022) | −0.086 (0.681) | −0.048 (0.327) |
| Random Forest | −0.015 (0.146) | 0.039 (0.030) | 0.018 (0.062) | −0.012 (0.116) | |
| Median R2 | |||||
| Blade 16 length | Elastic Net | 0.243 (0.002) | 0.322 (0.002) | 0.079 (0.002) | −0.060 (0.088) |
| Random Forest | 0.381 (0.002) | 0.473 (0.002) | 0.031 (0.012) | 0.017 (0.016) | |
| Blade 16 width | Elastic Net | 0.408 (0.002) | 0.232 (0.002) | 0.668 (0.002) | 0.537 (0.002) |
| Random Forest | 0.270 (0.002) | 0.321 (0.002) | 0.434 (0.002) | 0.292 (0.002) | |
| Husk leaf length | Elastic Net | 0.404 (0.002) | 0.435 (0.002) | 0.449 (0.002) | 0.443 (0.002) |
| Random Forest | 0.198 (0.002) | 0.233 (0.002) | 0.463 (0.002) | 0.446 (0.002) | |
| Ear length | Elastic Net | 0.376 (0.002) | 0.085 (0.002) | 0.370 (0.002) | 0.187 (0.004) |
| Random Forest | 0.009 (0.016) | −0.051 (0.044) | 0.312 (0.002) | 0.123 (0.002) | |
| Plant height | Elastic Net | −0.037 (0.048) | −0.291 (0.824) | −0.166 (0.399) | −0.106 (0.206) |
| Random Forest | −0.391 (0.858) | −0.323 (0.762) | −0.249 (0.467) | −0.081 (0.074) | |
| PCC | |||||
| Blade 16 length | Elastic Net | 0.587 (0.002) | 0.600 (0.002) | 0.603 (0.002) | 0.562 (0.002) |
| Random Forest | 0.606 (0.002) | 0.665 (0.002) | 0.566 (0.002) | 0.603 (0.002) | |
| Blade 16 width | Elastic Net | 0.717 (0.002) | 0.700 (0.002) | 0.805 (0.002) | 0.812 (0.002) |
| Random Forest | 0.647 (0.002) | 0.559 (0.002) | 0.753 (0.002) | 0.771 (0.002) | |
| Husk leaf length | Elastic Net | 0.677 (0.002) | 0.644 (0.002) | 0.718 (0.002) | 0.690 (0.002) |
| Random Forest | 0.568 (0.002) | 0.625 (0.002) | 0.757 (0.002) | 0.730 (0.002) | |
| Ear length | Elastic Net | 0.503 (0.002) | 0.385 (0.004) | 0.463 (0.004) | 0.535 (0.002) |
| Random Forest | 0.317 (0.006) | 0.265 (0.016) | 0.542 (0.002) | 0.372 (0.004) | |
| Plant height | Elastic Net | 0.268 (0.014) | 0.250 (0.012) | −0.016 (0.122) | −0.135 (0.224) |
| Random Forest | 0.115 (0.106) | 0.212 (0.030) | 0.172 (0.060) | 0.110 (0.098) | |
Three different sections of the table show the pooled R 2, median R 2, and Pearson correlation (PCC) measures for the prediction performance of the models learned for all traits using all transcripts (Transcripts), only regulatory transcripts (Regulators), all metabolites (Metabolites), and both transcripts and metabolites (Both) as features. Numbers between parentheses indicate P‐values for the performance values obtained, derived from permutation tests.