ChiENN: Embracing Molecular Chirality with Graph Neural Networks

Explore the innovative ChiENN (Chiral Edge Neural Network) approach for incorporating molecular chirality in Graph Neural Networks. Dive into the fundamentals of chirality in chemical compounds and its significance in drug discovery. Learn about the limitations of traditional GNNs in capturing chiral information and discover how ChiENN overcomes these challenges. Examine the theoretically justified message-passing scheme that enables GNNs to become sensitive to node neighbor order. Understand the application of this concept in molecular chirality and how ChiENN layers can be integrated into existing GNN models. Review experimental results demonstrating ChiENN's superior performance in chiral-sensitive molecular property prediction tasks compared to current state-of-the-art methods. Engage with the Q&A session to gain deeper insights into this cutting-edge research presented by Piotr Gaiński from Valence Labs.

- Intro
- Background
- Related Work
- Method
- Order-Sensitive Message Passing
- ChiENN
- Results
- Q&A