Table 10.
Golub cancer data set
| Ns | ||||
|---|---|---|---|---|
| 4 | 193 | 291 | ||
| I | ORdensity (A) | 90.28 | 97.22 | 97.22 |
| limma (B) | 97.22 | 97.22 | 97.22 | |
| SAM (C) | 97.22 | 97.22 | 97.22 | |
| II | A vs B | 1 | 127 | 206 |
| A vs C | 1 | 130 | 213 | |
| B vs C | 4 | 175 | 265 | |
| III | ORdensity | 0.5 (0.71) | 150.4 (3.89) | 241.8 (6.94) |
| limma | 0 | 12.3 (2.26) | 4.3 (1.83) | |
| SAM | 3.3 (0.48) | 151.8 (24.18) | 220.7 (35.88) | |
| Ns | ||||
| 403 | 556 | 938 | ||
| I | ORdensity (A) | 97.22 | 97.22 | - |
| limma (B) | 97.22 | 97.22 | 97.22 | |
| SAM (C) | 97.22 | - | - | |
| II | A vs B | 274 | 358 | - |
| A vs C | 280 | - | - | |
| B vs C | 368 | - | - | |
| III | ORdensity | 334.1 (6.03) | 473.3 (9.56) | - |
| limma | 029.2 (2.44) | 55.5 (3.10) | 149.3 (6.36) | |
| SAM | 316.3 (50.06) | - | - | |
Results for different number (Ns) of selected genes: 4 with ORdensity strong selection; 193 with limma and Bonferroni; 291 with ORdensity relaxed selection; 403 with SAM; 556 total potential DE genes and 938 with limma and BH. Rows I present the leave-one-out cross-validation correct classification rate. In bold, the results for the genes selected under the standard criteria for ORdensity, limma and SAM procedures; in rows II, number of common selected genes between the ORdensity, limma and SAM approaches; in rows III, mean and standard deviation (in brackets) of the number of genes that for 10 subsamples were always kept selected