-
Code for the paper: ChemGAN challenge for drug discovery: can AI reproduce natural chemical diversity? arXiv preprint arXiv:1708.08227.
-
Related blog post: https://medium.com/the-ai-lab/chemgan-challenge-for-drug-discovery-can-ai-reproduce-natural-chemical-diversity-8f1f2528ee22
-
Chat room: https://gitter.im/Startcrowd/drugdiscovery
-
Video by Siraj Raval: https://youtu.be/hY9Bc3mtphs
-
Requirements: Rdkit version 2017.03.3 from Anaconda, Tensorflow 1.0.1
-
The code has not been cleaned, don't hesitate to post an issue if you don't find what you are looking for.
-
To make the DRD2 case work, take clf.pkl here, rename it clf_drd2.pkl, and put it in the appropriate folder. It's the SVM activity model of DRD2 by Marcus Olivecrona.
-
To make the QED case work, you need Silicos-it.
-
In order to train the model, cd into
modeland run
python train_ogan.py exp.json
where exp.json is a experiment configuration file.
- This code is otherwise based on ORGAN. Many thanks to the ORGAN team.