Source code for the paper: De Novo Drug Design by Iterative Multi-Objective Deep Reinforcement Learning with graph-based Molecular Quality Assessment

Generating molecules of high quality and drug-likeness in the vast chemical space is a big challenge in the drug discovery. Most existing generating methods focus on diversity and novelty of molecules, but ignoring the drug potentials of the generated molecules during the generation process. In this study, we present a novel de novo multi-objective quality assessment-based drug design approach QADD, which integrates an iterative refinement framework with a novel graph-based molecule quality assessment model for drug potentials. QADD designs a multi-objective deep reinforcement learning pipeline to generate molecules with multiple desired properties iteratively, where a graph neural network-based model for accurate molecular quality assessment on drug potentials is introduced to guide molecule generation.

• python3.7
• torch 1.8.1
• skearn 0.21.3
• numpy 1.20.1
• dgl 0.6.1
• rdkit
• baselines

For the molecular quality assessment (QA) model (in the 'QA/' folder)

1. convert SMILES strings to graph by generating "graph.bin" file

   python dataQA.py 

2. train the QA model

   python trainQA.py

3. predict

   python predictQA.py
   

For the deep reinforcement learning (RL) model (in the 'RL/' folder)

python run_RL.py

For the iterative refinement: retrained QA model every 5000 episode

• All materials are made available under the terms of the Creative Commons Attribution-NonCommercial 4.0 International Public License (CC BY-NC 4.0) license. You can find details at: https://creativecommons.org/licenses/by-nc/4.0/legalcode
• The license gives permission for academic use only.

• [email protected]