RISE: Radius of Influence based Subgraph Extraction for 3D Molecular Graph Explanation

3D Geometric Graph Neural Networks (GNNs) have emerged as transformative tools for modeling molecular data. Despite their predictive power, these models often suffer from limited interpretability, raising concerns for scientific applications that require reliable and transparent insights. While existing methods have primarily focused on explaining molecular substructures in 2D GNNs, the transition to 3D GNNs introduces unique challenges, such as handling the implicit dense edge structures created by a cutoff radius. To tackle this, we introduce a novel explanation method specifically designed for 3D GNNs, which localizes the explanation to the immediate neighborhood of each node within the 3D space. Each node is assigned an radius of influence, defining the localized region within which message passing captures spatial and structural interactions crucial for the model's predictions. This method leverages the spatial and geometric characteristics inherent in 3D graphs. By constraining the subgraph to a localized radius of influence, the approach not only enhances interpretability but also aligns with the physical and structural dependencies typical of 3D graph applications, such as molecular learning.

Understanding how artificial intelligence (AI) models make decisions is crucial -- especially when they are used in chemistry or drug development. One promising type of molecular AI models, called 3D Graph Neural Networks (GNNs), is good at predicting molecular properties by considering the 3D shapes of molecules. But these models are often regarded as “black boxes” -- it’s hard to know how they make the final predictions.Our research introduces RISE (Radius of Influence-based Subgraph Extraction), a new pipeline designed to make these models more transparent and understandable. RISE works by identifying a “radius of influence” around each atom -- essentially a distance within which other atoms significantly affect the model’s prediction. By doing this, RISE can highlight the most important parts of a molecule in a way that aligns with real chemistry, such as identifying actual chemical bonds.Compared to previous methods, RISE is both more accurate and more interpretable. It avoids technical shortcuts about uninterpretable explanations, and it consistently identifies important subgraphs with less edges. This makes RISE especially useful for scientists who want to know how AI models make decisions for molecular learning tasks.