Reshaping RNA structural biology with AI and integrative cryo-mass spectrometry

In 2024, the Nobel Prize in Chemistry celebrated the power of artificial intelligence (AI) to predict protein structures with unprecedented accuracy. Yet RNA, which is arguably just as central to biology and medicine, remains structurally elusive. Why? Unlike proteins, RNA structure is highly dynamic and sensitive to functional context and thus often defies conventional structure determination and prediction methods. Today, less than 4% of available three-dimensional structures represent RNA, despite RNA making up the vast majority of our transcriptome. In this article, I explore how new technologies are changing this narrative. From my early doctoral work combining Nuclear Magnetic Resonance (NMR) spectroscopy with simulations to reveal hidden RNA intermediates to the development of a radically new approach using cryogenic Orbitrap secondary ion mass spectrometry (cryo-OrbiSIMS), we are now able to capture structural fingerprints of RNA directly from native samples. Building on this breakthrough, we are now integrating machine learning to predict RNA structures from cryo-OrbiSIMS spectral features. Setting the stage with this technological triad of cryo-mass spectrometry, structure modelling, and AI, we are beginning to unravel the structural complexity of RNA with speed, sensitivity, and scale to finally let RNA into the spotlight.