Geom-SAC: Geometric Multi-Discrete Soft Actor Critic With Applications in De Novo Drug Design

Code implementation of the paper: https://ieeexplore.ieee.org/document/10472486

Finding new molecules with desirable properties has high computational and overhead costs. Much research has focused on generating candidate molecules in one- and two-dimensional spaces which has produced some favorable results. However extending these approaches to molecules in three- dimensional space would be far more useful because the representation of molecules is more realistic although three-dimensional methods have much higher computational costs. In this work we developed a geometric deep reinforcement learning agent that generates and optimizes molecules that could interact with a biochemical target. The agent can be used for generating molecules from scratch or for lead optimization when it enhances the properties of a given molecule whether by enhancing its drug-likeness or increasing its activity toward the target via implicit learning. Thus the agent works with molecules in three-dimensional space without high computational costs

🧐 Features

Here're some of the project's best features:

• Scalable and efficient to large molecular systems.
• Integrates GNNs and DRL for drug discovery.
• Our algorithm obtains molecules with high QED values and learns the linking strategies that did not violate valency chemical validity or embeddability in 3D space.
• Modified Soft-Actor-Critic agent to handle multi-discrete actions space.

🛠️ Installation Steps:

1. TODO

TODO

🍰 Contribution Guidelines:

Opening issues:

Filing issues with the project is a great step in helping the project — you're helping us learn what's not working what you might want to see in the project or what we can do better. Even questions like 'how do I do X?' help us know where documentation may be lacking. When opening an issue please make sure to follow the issue templates. Doing that helps us gather the required initial information to be able to further assist.

Commenting

Adding meaningful comments to issues and pull requests helps the community as a whole. Providing suggestions helping solving issues giving feedback — it all helps. Once again the main guideline we have here is 'be kind'.

Voting

We're not big fans of adding comments like "+1" or "any updates on this?" but adding a 👍 reaction to the opening comment helps us track what attracts more interest in our community and what we should focus on next.

Code reviews

Even if you don't have time to contribute your own code you may have time to review existing pull requests helping us process those faster.