GCNBMP¶
- class GCNBMP(*, molecule_channels=69, num_relations=4, hidden_channels=16, hidden_conv_layers=1, out_channels=1)[source]¶
Bases:
chemicalx.models.base.Model
An implementation of the GCN-BMP model from [chen2020].
See also
This model was suggested in https://github.com/AstraZeneca/chemicalx/issues/21
- chen2020
Chen, X., et al. (2020). GCN-BMP: Investigating graph representation learning for DDI prediction task. Methods, 179, 47–54.
Methods Summary
forward
(molecules_left, molecules_right)Run a forward pass of the GCN-BMP model.
unpack
(batch)Return the left and right drugs PackedGraphs.
Methods Documentation
- forward(molecules_left, molecules_right)[source]¶
Run a forward pass of the GCN-BMP model.
- Parameters
molecules_left (
PackedGraph
) – The graph of left drug and node features.molecules_right (
PackedGraph
) – The graph of right drug and node features.
- Return type
FloatTensor
- Returns
A column vector of predicted synergy scores.