Table 2.
For the SGSR model, the top 7 predictive features are displayed for AZD6244 (AZD) and PD0325901(PD). Cells higlighted in orange correspond to features with evidence of being functionally related to MEK inhibitor compounds, as described in the text. For comparison, the ranks of corresponding predictive features inferred by bootstrapped Elastic Net are displayed (18,897 and 12,024 features are considered in model building with CCLE and Sanger, respectively).
| Biomarker | SGSR | Bootstrapped Elastic Net regression | ||||||
|---|---|---|---|---|---|---|---|---|
| CCLE | Sanger | CCLE | Sanger | |||||
| AZD | PD | AZD | PD | AZD | PD | AZD | PD | |
| EDIL3 | Y | Y | Y | Y | 7260 | 11170 | 5109 | 4361 |
| RHOA | Y | Y | Y | Y | 16833 | 18775 | 1335 | 5022 |
| FYN | NA | NA | Y | Y | 11962 | 9574 | 7932 | 10345 |
| MAPK1 | Y | Y | NA | Y | 618 | 815 | 8914 | 10352 |
| MAP2K1 | Y | Y | NA | Y | 10133 | 11896 | 11924 | 8338 |
| AURKB | Y | Y | Y | Y | 12979 | 16464 | 6772 | 8464 |
| SRC | NA | NA | Y | Y | 10820 | 16675 | 6501 | 8516 |