tdc data_process

MOSES

wget https://media.githubusercontent.com/media/molecularsets/moses/master/data/dataset_v1.csv 

mv dataset_v1.csv raw_data/moses.csv

python data_process/moses.py 

[Paired data]

wget https://raw.githubusercontent.com/wengong-jin/iclr19-graph2graph/master/data/drd2/train_pairs.txt

wget https://raw.githubusercontent.com/wengong-jin/iclr19-graph2graph/master/data/qed/train_pairs.txt

wget https://raw.githubusercontent.com/wengong-jin/iclr19-graph2graph/master/data/logp04/train_pairs.txt

ZINC

pls see https://github.com/microsoft/constrained-graph-variational-autoencoder/blob/master/data/get_zinc.py

wget https://raw.githubusercontent.com/aspuru-guzik-group/chemical_vae/master/models/zinc_properties/250k_rndm_zinc_drugs_clean_3.csv 

rxn_yields

It contains 2 datasets: (1) Buchwald-Hartwig and (2)Suzuki-Miyaura

## update training_scripts/launch_buchwald_hartwig_training.py 
cd rxn_yields/training_scripts
cd /Users/futianfan/Downloads/summer_2020/rxn_yields/training_scripts
python launch_buchwald_hartwig_training.py finetuned 
## setup env for rxn_yields is quite complex, so it is not introduced here. 
## I only modify launch_buchwald_hartwig_training.py and copy it into data_process repo -> output is "buchwald.csv", in raw_data



python data_process/buchwald_yield.py