Table 6. Comparison of BioAct-Het While Compounds Are Represented by the GCN-AttentiveFP Model and Jiang et al. Model Based on the Association Strategy.
| data set | model name | algorithm | AUC-ROC (%) |
|---|---|---|---|
| SIDER | Jiang et al. model | RF | 64.6 |
| XGBoost | 64.2 | ||
| DNN | 63.1 | ||
| SVM | 63 | ||
| MPNN | 59.8 | ||
| GCN | 63.4 | ||
| GAT | 62.7 | ||
| AttentiveFP | 62.3 | ||
| BioAct-Het | association-based | 91.11 | |
| Tox21 | Jiang et al. model | RF | 83.8 |
| XGBoost | 83.6 | ||
| DNN | 84 | ||
| SVM | 81.7 | ||
| MPNN | 80.9 | ||
| GCN | 83.6 | ||
| GAT | 83.5 | ||
| AttentiveFP | 85.2 | ||
| BioAct-Het | association-based | 89.80 | |
| MUV | Jiang et al. model | RF | 6.1 |
| XGBoost | 6.8 | ||
| DNN | 2.1 | ||
| SVM | 11.2 | ||
| MPNN | 1.6 | ||
| GCN | 6.1 | ||
| GAT | 5.7 | ||
| AttentiveFP | 3.8 | ||
| BioAct-Het | association-based | 69.47 |