Designing Random Graph Models Using Variational Autoencoders With Applications to Chemical Design

Paper : https://arxiv.org/pdf/1802.05283.pdf

Required packages

-tensorflow 1.4.1

-rdkit >= 2016.03.4

-networkx 2.0

Command to learn model

python main.py --num_epochs 10 --learning_rate 0.0001 --log_every 5 --z_dim <z> --random_walk <k> --edges <e> --nodes <n> --graph_file <graph> --z_dir <zspace> --sample_file <sampledir> --out_dir <outputdir> > log.out

Command to sample graph

python sample.py --num_epochs 10 --learning_rate 0.0001 --log_every 5 --z_dim <z> --random_walk <k> --edges <e> --nodes <n> --graph_file <graph> --z_dir <zspace> --sample_file <sample> --out_dir <outputfile> > log.out

For real data please checkout the branch node_label