Computational Methods for Assembling Full-Length Circular RNAs

This research project aims to develop new computer methods to accurately find and assemble circular RNAs (circRNAs) from different types of RNA sequencing data. Unlike the regular RNA molecules with two ends, circRNAs are special molecules in cells that form a closed loop, making them highly resistant to degradation and allowing them to be present in the cells for a much longer time. They are known to play important roles in how genes are expressed in response to stress or a condition and can be used as biomarkers to identify a cellular condition or response. However, accurately identifying complete circRNA molecules from modern RNA sequencing data remains a major computational challenge due to their low abundance, circular structure, and the limitations of current sequencing technologies. This project addresses these fundamental challenges by creating innovative methods that improve the accuracy and reliability of circRNA reconstruction from RNA sequencing data from diverse experiments and sequencing platforms. These tools will help scientists discover new circRNAs, understand their functions, and possibly use them as biomarkers. Educational activities include training undergraduate and graduate students; engaging K-12 teachers through Penn State's Research Experiences for Teachers program; and developing high-school curriculum materials and disseminating them. This project plans to develop a unified computational framework for assembling full-length circRNAs from short-read and long-read RNA-seq data. The research will pursue four integrated objectives. First, it will develop highly sensitive algorithms that exploit previously underutilized sequencing signals to improve the circRNA assembly from the RNA sequencing data. Second, it will develop novel reference-free algorithms for reconstructing circRNAs from rolling circular long reads by jointly assembling multiple reads from the same molecule. Third, it will create reference-based algorithms for long-read sequencing data, addressing sequencing errors and assembly ambiguities through graph-based optimization techniques. Fourth, the project will develop machine-learning methods to score assembled circRNAs and statistical methods for false discovery rate control, enabling researchers to prioritize high-confidence predictions with rigorous statistical guarantees. The computational methods and the resulting circRNA predictions will be rigorously validated through comprehensive computational and experimental studies. Together, these advances will contribute to biotechnology innovation and the development of artificial intelligence applications for biology. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.