TABLE 2.
Disease classification results of SVM classifiers using 15 subnetwork markers and other known gene signatures.
| Accuracy (%) | Sensitivity | Specificity | AUC | |
|---|---|---|---|---|
| 15 Subnetworks | 98 | 0.98 | 1.00 | 0.99 |
| 24 DEGs* | 97 | 0.94 | 1.00 | 0.97 |
| ColoGuideEx | 84 | 0.83 | 1.00 | 0.91 |
| ColoPrint | 84 | 0.82 | 1.00 | 0.91 |
| Genefx | 98 | 0.98 | 1.00 | 0.99 |
| Oncotype DX | 87 | 0.85 | 1.00 | 0.93 |
Abbreviations: AUC, Area Under Curve; DEG, Differentially expressed genes. *DEGs with copy number alterations identified in this study (Figure 1B). All classifiers for gene signatures, ColoGuideEx (59), ColoPrint (60), Genefx (61), and Oncotype DX (62) and 24 DEGs, are designed using GSE23878 dataset as training and TCGA dataset as validation.