The next wave of RNA (ribonucleic acid) engineering will be driven by AI agents that select the most suitable tools, generate candidate RNA sequences, and produce calibrated quality metrics to guide experiments. Over the past few decades, a variety of computational methods for RNA folding and design have been developed, along with new benchmark datasets. Each of them is optimised for different assumptions, objective functions, and constraints. This poses a challenge for experimentalists who may lack computational expertise, as they may struggle to choose the right tools for a specific RNA design. Can we have a unified platform that surveys the existing database of RNA folding/design tools, selects the appropriate method, runs it, and suggests designed RNA sequences based on the experimentalist's specifications? With the variety of RNA folding and design tools developed over the past few decades and new benchmark datasets—each optimised for different assumptions, objectives, and constraints—it is surely challenging to choose the right tool for a specific RNA design task. This is especially true for experimentalists who may lack extensive computational expertise. This project proposes to develop a user-friendly RNA design platform based on a Random Forest decision model. Given a set of RNA target structures and a set of design constraints, our decision model will recommend appropriate folding and design tools to generate candidate RNA sequences along with quality metrics to guide experimental validation.
This project received an approximate amount of 20 000 € from the ScaDS.AI Early Career Innovation Projects on the Future of AI for a proof of concept implementation.