Table 4 Performances of different molecular training tasks
From: SubMDTA: drug target affinity prediction based on substructure extraction and multi-scale features
Model | MSE | CI | \(r_m^2\) |
|---|---|---|---|
SubMDTA-a(no pretraining) | 0.224 | 0.894 | 0.714 |
SubMDTA-b (only subgraph-level pretraining) | 0.225 | 0.897 | 0.703 |
SubMDTA-c (only graph-level pretraining) | 0.230 | 0.895 | 0.697 |
SubMDTA (both subgraph-level and graph-level pretraining) | 0.218 | 0.894 | 0.719 |