HARMONY: A Multi-Representation Framework for RNA Property Prediction

The biological functions of RNA arise from the interplay of sequence (1D), secondary structure (2D), and tertiary structure (3D). While existing machine learning models typically rely on sequence-based representations, recent studies suggest that integrating structural information can improve predictive performance, especially in low-data regimes. However, different representations have trade-offs—3D models are sensitive to noise, whereas sequence-based models are more robust to sequencing noise but lack structural insights. To address this, we introduce HARMONY, a framework that dynamically integrates 1D, 2D, and 3D representations, and seamlessly adapts to diverse real-world scenarios. Our experiments demonstrate that HARMONY consistently outperforms existing baselines across multiple RNA property prediction tasks on established benchmarks, offering a robust and generalizable approach to RNA modeling.