Table 1.
The dimensions and parameters of DeepPROTACs model
| DeepPROTACs model | Layer | Output dimension |
|---|---|---|
| Protein pockets/ligands | Node embedding | 64 |
| Graph convolution layer 1 | 128 | |
| Graph convolution layer 2 | 64 | |
| SMILES | Embedding | 64 |
| Bidirectional LSTM | 64 | |
| Fully connected layer | 64 | |
| MLP | Fully connected layer 1 | 64 |
| Fully connected layer 2 | 2 | |
| Parameters | Epoch | 30 |
| Batch size | 1 | |
| Number of GCN layers | 2 | |
| Pocket size | 5 Å | |
| Inclusion of bond type encoding | Yes | |
| Learning rate | 0.0001 | |
| Pooling layer | Max pooling | |
| Loss | Cross-entropy | |
| Optimizer | Adam |