Improving Performance Prediction of Synthetic RNA Switches with High-Throughput Screening and Deep Learning

• Applicant: Dr. Tianhe Wang, Ph.D.
• Subject Area: Biochemistry
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 566109210

RNA aptamers and engineered riboswitches have emerged as versatile programmable tools with diverse applications in clinical diagnostics, therapeutics, and synthetic biology. However, engineering and optimizing the regulatory behavior of these synthetic RNA components remains a challenging task, often involving a complex iterative process. This situation presents an opportunity to leverage advanced pattern recognition capabilities offered by deep learning (DL) techniques. The overarching goal of this project is to improve the switching performance of aptamer-based RNA switches both by rational design and screening in vitro and in vivo, while simultaneously utilizing the resulting dataset to fine-tune the existing DL models for predictive analysis. The objective of this proposal is to initially employ directed evolution approaches, including Capture-SELEX, fluorescence-dependent droplet and cell sorting-sequencing to screen and characterize variant libraries of RNA switches. The switching behavior of these variants will be evaluated based on sequencing data, secondary structure and reporter gene analyses. It will lead to improved switching behavior, but also populate a comprehensive database that, subsequently, will then be utilized to fine-tune pre-trained DL models to realize the prediction of improved aptamer-based RNA switches. Ultimately, this 'Design-Build-Test+Learn' workflow aims to deliver high-performance RNA switches alongside efficient predictive models for the future switch design, potentially leading to significant improvements in the performance of RNA devices in various synthetic biology applications.

• DFG Programme: WBP Position