Graph deep learning locates magnesium ions in RNA.

Magnesium ions (Mg) are vital for RNA structure and cellular functions. Present efforts in RNA structure determination and understanding of RNA functions are hampered by the inability to accurately locate Mg ions in an RNA. Here we present a machine-learning method, originally developed for computer visual recognition, to predict Mg binding sites in RNA molecules. By incorporating geometrical and electrostatic features of RNA, we capture the key ingredients of Mg-RNA interactions, and from deep learning, predict the Mg density distribution. Five-fold cross-validation on a dataset of 177 selected Mg-containing structures and comparisons with different methods validate the approach. This new approach predicts Mg binding sites with notably higher accuracy and efficiency. More importantly, saliency analysis for eight different Mg binding motifs indicates that the model can reveal critical coordinating atoms for Mg ions and ion-RNA inner/outer-sphere coordination. Furthermore, implementation of the model uncovers new Mg binding motifs. This new approach may be combined with X-ray crystallography structure determination to pinpoint the metal ion binding sites.