New methods for computational modeling of RNA structures

University Of Missouri-Columbia, Columbia MO

Investigators

Linked publications & trials

Abstract

Project Summary The past five years have ushered in a transformative era in biomedical research. With the development of life- saving SARS-CoV-2 mRNA vaccines, FDA approval of the first RNA-targeted drug, and FDA approval of the first CRISPR-based gene editing therapy, RNA is revolutionizing key areas of medicine. Despite these remarkable advances, the full potential of these technologies remains untapped, partly due to bottlenecks in quantitative RNA modeling and design. While machine learning (ML) has profoundly influenced biomolecular modeling, it has yet to capture the complexities of RNA's heterogeneous conformational distributions, alternative folds, and kinetics — these are the hallmarks of RNA structure and function. Recent advancements in ML and the ever- growing data from RNA-based technologies have now positioned us to develop new biophysical tools and make significant breakthroughs in understanding, modeling, and designing RNAs. Leveraging the extensive structural and biophysical data, including over 500,000 thermodynamic and structural probing data points for more than 100,000 designed RNA sequences from experimental collaborators, we will develop hybrid approaches that integrate cutting-edge machine learning techniques with biophysical methods. We aim to develop tools targeting significant problems in RNA biology — from the prediction of RNA conformational distributions, RNA-small molecule interactions, and cotranscriptional folding to the learning and large-scale modeling of structure-based mRNA vaccine design. The new tools developed in this grant will provide a foundation for future designs of mRNA vaccines, antiviral drugs, and CRISPR genome editing.