From: SSF-DDI: a deep learning method utilizing drug sequence and substructure features for drug–drug interaction prediction
Layer name
Output size
Parameters
DMPNN
Atom feature dim 64
In_features 70, out_features 64, hidden dim 64, n_iter 10
GATConv
Atom feature dim 70
In_channels 70, out_channels 35, heads 2
SAGPooling
[batch_size,64]
In_features 64,out_features 1