Awesome

FusionRetro: Molecule Representation Fusion via In-Context Learning for Retrosynthetic Planning

This repository contains an implementation of "FusionRetro: Molecule Representation Fusion via In-Context Learning for Retrosynthetic Planning", which is an autoregressive framework for molecule synthetic route generation.

Warning

We only use one route for each molecule in the training dataset for model training!!!

Contribution

Zuobai Zhang contributes the implementation of G2Gs, while I contribute the rest.

Dropbox

We provide the starting material file in dropbox, you can download this file via: https://www.dropbox.com/scl/fi/j3kh641irxtpbrnjnmoop/zinc_stock_17_04_20.hdf5?rlkey=zqbymj13skpdqlswu2uvji1sq&st=c1805gz0&dl=0 Please move this file into the root folder.

FusionRetro

cp train_dataset.json valid_dataset.json test_dataset.json zinc_stock_17_04_20.hdf5 FusionRetro/  
cd FusionRetro  

#Data Process  
python to_canolize.py --dataset train  
python to_canolize.py --dataset valid  
python to_canolize.py --dataset test  

#Initial Train
python train.py --batch_size 64 --epochs 3000  
# After 3000 epochs, We set global_step to 1000000 and continue to train the model (3000th epoch's model paramater) with 1000 epochs  
#Continue Train
python train.py --batch_size 64 --continue_train --epochs 1000

# We select the model with the performance on the first 100 routes in the validation dataset

#We also provide model.pkl, you can skip the above commands

#Retro Star Zero Search
python retro_star_0.py  --beam_size 5  

#Retro Star Search
python get_reaction_cost.py  
python get_molecule_cost.py  
python value_mlp.py
#We also provide value_mlp.pkl, you can skip the above commands
python retro_star.py --beam_size 5  

#Greedy DFS Search
python greedy_dfs.py --beam_size 5  

Transformer

cp train_dataset.json valid_dataset.json test_dataset.json zinc_stock_17_04_20.hdf5 Transformer/  
cd Transformer  

#Data Process  
python to_canolize.py --dataset train  
python to_canolize.py --dataset valid  
python to_canolize.py --dataset test  

#Train  
python train.py --batch_size 32 --epochs 2000  

# We select the model with the performance on the first 100 routes in the validation dataset

#We also provide model.pkl, you can skip the above commands

#Retrosynthesis Test
python retrosynthesis_test.py --beam_size 10  

#Retro Star Zero Search
python retro_star_0.py  --beam_size 5  

#Retro Star Search
python get_reaction_cost.py  
python get_molecule_cost.py  
python value_mlp.py
#We also provide value_mlp.pkl, you can skip the above commands
python retro_star.py --beam_size 5  

#Greedy DFS Search
python greedy_dfs.py --beam_size 5  

Retrosim

cp train_dataset.json valid_dataset.json test_dataset.json zinc_stock_17_04_20.hdf5 Retrosim/  
cd Retrosim  

#Retrosynthesis Test
python retrosynthesis_test.py --beam_size 10 --num_cores 64  

#Retro Star Zero Search
python retro_star_0.py  --beam_size 5 --num_cores 64  

#Retro Star Search
python get_reaction_cost.py  
python get_molecule_cost.py  
python value_mlp.py
#We also provide value_mlp.pkl, you can skip the above commands
python retro_star.py --beam_size 5 --num_cores 64  

#Greedy DFS Search
python greedy_dfs.py --beam_size 5 --num_cores 64  

Neuralsym

cp train_dataset.json valid_dataset.json test_dataset.json zinc_stock_17_04_20.hdf5 Neuralsym/  
cd Neuralsym  

#Data Process
python prepare_data.py  

#Train
bash train.sh  

# We select the model by the original' code's setting 

#Retrosynthesis Test
python retrosynthesis_test.py --beam_size 10 --num_cores 64  

#Retro Star Zero Search
python retro_star_0.py  --beam_size 5 --num_cores 64  

#Retro Star Search
python get_reaction_cost.py  
python get_molecule_cost.py  
python value_mlp.py
#We also provide value_mlp.pkl, you can skip the above commands
python retro_star.py --beam_size 5 --num_cores 64  

#Greedy DFS Search
python greedy_dfs.py --beam_size 5 --num_cores 64  

GLN

cp train_dataset.json valid_dataset.json test_dataset.json zinc_stock_17_04_20.hdf5 GLN/gln/  
cd GLN  
pip install -e .  
cd gln  

#Data Process 
python process_data_stage_1.py -save_dir data  

python process_data_stage_2.py -save_dir data -num_cores 12 -num_parts 1 -fp_degree 2 -f_atoms data/atom_list.txt -retro_during_train False $@  

python process_data_stage_2.py -save_dir data -num_cores 12 -num_parts 1 -fp_degree 2 -f_atoms data/atom_list.txt -retro_during_train True $@  

#Train
bash run_mf.sh schneider  

# We select the model with the performance on all routes in the validation dataset

#Retrosynthesis Test
python retrosynthesis_test.py -save_dir data -f_atoms data/atom_list.txt -gpu 0 -seed 42 -beam_size 10 -epoch_for_test 100  

#Retro Star Zero Search
python retro_star_0.py -save_dir data -f_atoms data/atom_list.txt -gpu 0 -seed 42 -beam_size 5 -epoch_for_search 100  

#Retro Star Search
python get_reaction_cost.py -save_dir data -f_atoms data/atom_list.txt -gpu 0 -seed 42 -beam_size 10 -epoch_for_search 100  
python get_molecule_cost.py  
python value_mlp.py
#We also provide value_mlp.pkl, you can skip the above commands
python retro_star.py -save_dir data -f_atoms data/atom_list.txt -gpu 0 -seed 42 -beam_size 5 -epoch_for_search 100  

#Greedy DFS Search
python greedy_dfs.py -save_dir data -f_atoms data/atom_list.txt -gpu 0 -seed 42 -beam_size 5 -epoch_for_search 100  

Megan

cp train_dataset.json valid_dataset.json test_dataset.json zinc_stock_17_04_20.hdf5 Megan/data/  
mv Megan/data/valid_dataset.json Megan/data/val_dataset.json  
 
cd Megan  
source env.sh  

#Data Process  
python json2csv.py  
python acquire.py uspto_50k  
python featurize.py uspto_50k megan_16_bfs_randat  

#Train
python bin/train.py uspto_50k models/uspto_50k  

# We select the model by the original' code's setting  

#Retrosynthesis Test
python bin/retrosynthesis_test.py models/uspto_50k --beam-size 10  

#Retro Star Search
python bin/get_reaction_cost.py models/uspto_50k --beam-size 10  
python bin/get_molecule_cost.py  
python bin/value_mlp.py
#We also provide value_mlp.pkl, you can skip the above commands
python bin/retro_star.py models/uspto_50k --beam-size 5  

#Retro Star Zero Search
python bin/retro_star_0.py models/uspto_50k --beam-size 5  

#Greedy DFS Search
python bin/greedy_dfs.py models/uspto_50k --beam-size 5  

GraphRetro

cp train_dataset.json valid_dataset.json test_dataset.json zinc_stock_17_04_20.hdf5 GraphRetro/datasets/uspto-50k  
cd GraphRetro  
export SEQ_GRAPH_RETRO=$(pwd)  
python setup.py develop  

#Data Process
mv datasets/uspto-50k/valid_dataset.json datasets/uspto-50k/eval_dataset.json  
python json2csv.py  
python data_process/canonicalize_prod.py --filename train.csv  
python data_process/canonicalize_prod.py --filename eval.csv  
python data_process/canonicalize_prod.py --filename test.csv  
python data_process/parse_info.py --mode train  
python data_process/parse_info.py --mode eval  
python data_process/parse_info.py --mode test  
python data_process/core_edits/bond_edits.py  
python data_process/lg_edits/lg_classifier.py  
python data_process/lg_edits/lg_tensors.py  

#Train
python scripts/benchmarks/run_model.py --config_file configs/single_edit/defaults.yaml  
python scripts/benchmarks/run_model.py --config_file configs/lg_ind/defaults.yaml  

# We select the model by the original' code's setting

#We also provide model files, you can skip the above commands

#Retrosynthesis Test
python scripts/eval/retrosynthesis_test.py --beam_size 10 --edits_exp SingleEdit_20220823_044246 --lg_exp LGIndEmbed_20220823_04432 --edits_step best_model --lg_step best_model --exp_dir models  

#Retro Star Search
python scripts/eval/get_reaction_cost.py --beam_size 10 --edits_exp SingleEdit_20220823_044246 --lg_exp LGIndEmbed_20220823_04432 --edits_step best_model --lg_step best_model --exp_dir models  
python scripts/eval/get_molecule_cost.py  
python scripts/eval/value_mlp.py
#We also provide value_mlp.pkl, you can skip the above commands
python scripts/eval/retro_star.py --beam_size 5 --edits_exp SingleEdit_20220823_044246 --lg_exp LGIndEmbed_20220823_04432 --edits_step best_model --lg_step best_model --exp_dir models  

#Retro Star Zero Search
python scripts/eval/retro_star_0.py --beam_size 5 --edits_exp SingleEdit_20220823_044246 --lg_exp LGIndEmbed_20220823_04432 --edits_step best_model --lg_step best_model --exp_dir models  

#Retrosynthesis Test
python scripts/eval/greedy_dfs.py --beam_size 5 --edits_exp SingleEdit_20220823_044246 --lg_exp LGIndEmbed_20220823_04432 --edits_step best_model --lg_step best_model --exp_dir models  

G2Gs

cp train_dataset.json valid_dataset.json test_dataset.json zinc_stock_17_04_20.hdf5 G2Gs/datasets/  
cd G2Gs  

#Train
python script/train.py -g [0]  

# We select the model by the original' code's setting  

#Retrosynthesis Test
python script/retrosynthesis_test.py -g [0] -k 10 -b 1  

#Retro Star Search
python script/get_reaction_cost.py -g [0] -k 10 -b 1  
python get_molecule_cost.py  
python value_mlp.py
#We also provide value_mlp.pkl, you can skip the above commands
python script/retro_star.py -g [0] -k 5 -b 1  

#Retro Star Zero Search
python script/retro_star_0.py -g [0] -k 5 -b 1  

#Greedy DFS Search
python script/greedy_dfs.py -g [0] -k 5 -b 1  

Acknowledgement

My deepest thanks to Binghong Chen and Samuel Genheden for very helpful discussions on their benchmarks (Retro* and PaRoutes)!

Reference

Retrosim: https://github.com/connorcoley/retrosim
Neuralsym: https://github.com/linminhtoo/neuralsym
GLN: https://github.com/Hanjun-Dai/GLN
G2Gs: https://torchdrug.ai/docs/tutorials/retrosynthesis
GraphRetro: https://github.com/vsomnath/graphretro
Transformer: https://github.com/bigchem/synthesis
Megan: https://github.com/molecule-one/megan

Citation

@inproceedings{liu2023fusionretro,
  title={FusionRetro: Molecule Representation Fusion via In-Context Learning for Retrosynthetic Planning},
  author={Liu, Songtao and Tu, Zhengkai and Xu, Minkai and Zhang, Zuobai and Lin, Lu and Ying, Rex and Tang, Jian and Zhao, Peilin and Wu, Dinghao},
  booktitle={International Conference on Machine Learning},
  year={2023}
}