A combined NMR and deep neural network approach for enhancing the spectral resolution of aromatic side chains in proteins @ScienceAdvances 1. This study introduces FID-Net-2, a deep neural network designed to enhance the spectral resolution of aromatic side chains in proteins by transforming standard NMR data into high-resolution 1H-13C correlation maps. 2. By combining advanced NMR experimental designs and machine learning, the method overcomes traditional challenges like multiplet splitting and signal attenuation, significantly improving spectral quality for aromatic residues. 3. FID-Net-2 was trained on synthetic datasets simulating diverse NMR conditions, ensuring robust performance and providing accurate uncertainty estimates for reconstructed spectra. 4. The methodology enables high-resolution mapping of aromatic side chains in medium- to large-sized proteins, such as a 40-kDa maltose-binding protein, which are typically challenging with conventional techniques. 5. The approach demonstrates utility in assigning chemical shifts, analyzing molecular dynamics, and characterizing chemical exchange processes with quantitative accuracy. 6. The development highlights a new paradigm in NMR spectroscopy by tailoring experimental designs specifically for deep learning applications, setting the stage for broader AI-assisted advancements. @DFlemmingHansen 💻Code: github.com/gogulan-k/FID-Net 📜Paper: science.org/doi/10.1126/scia… #NMR #DeepLearning #StructuralBiology #Bioinformatics #ProteinDynamics