Stochastic-based Generative Network Complex
A python package for the Stochastic-based Generative Network Complex
-
Submit thousands of jobs to generate latent space vectors.
cd sbatch python submit_generator.py 20221209
-
Divide generated latent space vectors to sub-files. Each sub-file has 2000 records.
cd .. python ./utils/divide_generated_ls.py 20221209
-
Decode all generated latent space vectors to smiles.
cd sbatch python submit_decode.py 20221209
-
Drop duplicated and unlikely smiles.
cd .. python ./utils/drop_duplicates.py 20221209
-
Encode generated smiles to latent space vectors
cd sbatch python submit_encoder.py 20221209
-
Binding affinity test
cd .. python ./src/filtered.py --date 20221209
-
ADMET and SAS test
Test ADMET on a online server: ADMET and download a csv file. Then transfer this file to server
scp ADMET.csv wangru25@hpcc.msu.edu:/mnt/research/guowei-search.8/RuiWang/FokkerPlanckAutoEncoder/results/generator_20221209
Then check if there is a molecule that falls in the optimal range.
python ./src/properties.py 20221209
- Decode latent space vectors (from encoder) to smiles.
- Make comparasion with the generated smiles. Check the reproduction rate.
[1] Wang, R., Feng, H. and Wei, G.W., 2023. ChatGPT in Drug Discovery: A Case Study on Anticocaine Addiction Drug Development with Chatbots. Journal of Chemical Information and Modeling, 63(22), pp.7189-7209.