I’m thinking of framework for graph-based deep lerning, and interested in molecular generation. I think one of the famouns frameworks for the graph-based deep lerning “pytorch geometric (PyG)”, and I also found that DeepChem exsits.
Would you advice me what advantages are in DeepChem compared to PyG?
I’d like to test molecular generation in near future.
I’d like to use three kinds of features: global, node, edge.
I have interest in Flow-based generation.
i have interset in 3d-graph generation, not SMILES based generation based on LSTM, etc.