Table 3 Performance of proposed and baseline models in task of drug-blind prediction.
From: Drug discovery and mechanism prediction with explainable graph neural networks
Method | Conv type | RMSE (\(\downarrow\)) | PCC (\(\uparrow\)) | \(\mathbf {R^2}\) (\(\uparrow\)) |
|---|---|---|---|---|
XGDP (w/o attn) | GCN | 0.045 ± 0.018 | 0.457 ± 0.476 | 0.480 ± 0.416 |
GAT | 0.038 ± 0.000 | 0.831 ± 0.003 | 0.679 ± 0.000 | |
GAT_E | 0.037 ± 0.001 | 0.834 ± 0.010 | 0.691 ± 0.020 | |
GATv2_E | 0.035 ± 0.000 | 0.847 ± 0.001 | 0.718 ± 0.002 | |
XGDP | GCN | 0.026 ± 0.000 | 0.918 ± 0.001 | 0.843 ± 0.002 |
GAT | 0.026 ± 0.000 | 0.923 ± 0.000 | 0.851 ± 0.001 | |
GAT_E | 0.026 ± 0.000 | 0.922 ± 0.001 | 0.849 ± 0.001 | |
GATv2_E | 0.026 ± 0.000 | 0.921 ± 0.001 | 0.846 ± 0.001 |
