Computational models of RNA structures are instrumental in elucidating RNA folding, function, and interactions. Such models make predictions of RNA secondary and tertiary structure based on multiple algorithms employing thermodynamics, comparative sequence analysis, and machine learning. Vienna RNA, RNA fold, and Mfold utilize minimum free energy (MFE) approaches for secondary structure prediction, whereas more sophisticated models such as Rosetta and FARFAR2 employ molecular dynamics and Monte Carlo simulations to predict tertiary structures. Such models are crucial in applications ranging from RNA therapeutics to riboswitch engineering and synthetic biology.
RNA structure prediction has become much more advanced through the use of deep learning and artificial intelligence-based methods such as EternaFold and AlphaFold-RNA. These models combine high-scale biological data to improve precision in structural modeling. Computational methods also assist in deciphering RNA-protein interactions and RNA-drug binding processes, which are essential for drug discovery. Challenges persist, however, in precisely modeling large RNAs as well as in the capture of dynamic conformational alterations, which calls for ongoing development of hybrid experimental-computational methods for improved RNA structure prediction.