RNA biology has entered a transformative era. Advances in single-cell and spatial transcriptomics, long-read sequencing, and high-throughput assays are generating data at unprecedented resolution and scale. At the same time, powerful deep learning approaches—particularly large language and foundation models—are rapidly redefining what is possible in extracting biological insight from these multidimensional datasets. Together, these developments open extraordinary opportunities for understanding RNA regulation in health and disease, while also posing urgent computational and conceptual challenges.

RNA molecules play central roles in regulating gene expression, yet their structures, modifications, and interactions remain only partly understood. They fold into complex three-dimensional shapes, form dynamic ribonucleoprotein assemblies with proteins and other RNAs, and undergo diverse chemical modifications that fine-tune their activity. Each of these layers adds regulatory potential but also complicates our ability to model and predict RNA behavior. Despite recent progress, we still lack a comprehensive view of how RNA sequence, structure, and context combine to drive function. Capturing this complexity requires not only sophisticated algorithms but also carefully designed experiments. New assays reveal transcriptome-wide RNA structures, modifications, localization, and dynamics with unprecedented detail, but translating these measurements into biological understanding demands close collaboration with AI specialists. Likewise, computational models require experimental validation and new types of data to reach their full potential. This interdependence makes the dialogue between communities essential.

This Dagstuhl Seminar will bring together computer scientists, computational biologists, and experimental researchers to jointly explore these frontiers. Four themes will guide our discussions:

1. Learning the language of RNA – foundation models, sequence-to-function prediction, and generative design for therapeutics.
2. Modeling interactions and structure – capturing intra- and intermolecular interactions of RNAs and their partners at high resolution.
3. The epitranscriptome – chemical modifications as a hidden layer of regulation, and the role of AI in deciphering them.
4. RNA dynamics in time and space – integrating single-cell, long-read, and spatial technologies to follow RNA life cycles in real biological contexts.

By fostering exchange across disciplines, the seminar will create a forum not only to assess current methods but also to define benchmarks, identify critical datasets, and set research priorities. We aim to accelerate the development of interpretable, scalable, and biologically grounded models—tools that will transform our understanding of gene regulation and enable the next generation of RNA-based therapies.

Creative Commons BY 4.0

Hani Goodarzi, Annalisa Marsico, Igor Ulitsky, and Kathi Zarnack