Predicted and Explained: Transforming drug discovery with AI for high-precision receptor-ligand interaction modeling...

The pharmaceutical industry faces persistent challenges in developing effective treatments for complex diseases, creating an urgent need for innovative approaches to accelerate drug discovery. A pivotal factor in this process is the accurate prediction of receptor-ligand interactions, which are critical for ensuring drug efficacy and safety. Traditional prediction methods are often time-consuming and resource-intensive, necessitating more efficient computational approaches. In this study, we present a machine learning framework that accurately predicts docking scores by integrating three complementary molecular representations: Lipinski descriptors, fingerprints, and graph-based representations. To enhance predictive performance, we proposed two fusion strategies: early fusion (feature-level integration) and late fusion (decision-level aggregation). Notably, the early fusion model outperformed other approaches, demonstrating that combining diverse molecular representations enhances both predictive accuracy and robustness. To further improve interpretability, we applied Local Interpretable Model-agnostic Explanations method, which identified critical physicochemical and structural features driving predictions. This approach clarified binding dynamics by illustrating how specific features influence docking scores. Validation against established bioinformatics tools and 3D visualizations confirmed the framework's biological plausibility and reliability. By unifying multi-scale molecular representations with advanced machine learning techniques, our framework not only enhances prediction accuracy but also provides insights into ligand-receptor binding mechanisms. This advancement accelerates therapeutic development by enabling faster, data-driven prioritization of candidate molecules for complex diseases.