Artificial Intelligence enhances the exploration of RNA structure and metabolism.
By Namira Hossain
AI is a technological revolution that is drastically altering the state of the world. It has shown to be a powerful ally in figuring out the intricate relationships between RNA structure and metabolism, opening up new directions for scientific study. Within living cells, the transmission of genetic information is fundamentally mediated by RNA. Because of its vital roles in protein synthesis, regulation of genes, and cellular interaction, RNA has taken the attention off of its more well-known cousin, DNA, which had previously dominated the limelight. Understanding the complex folds and connections found in RNA molecules, nevertheless, has long been a difficult task for researchers.
The secrets of RNA structure and metabolism are being solved using machine learning methods, a branch of artificial intelligence. Large datasets may be analyzed, RNA structure predictions can be made, and possible molecular connections can be found using these methods. Historically, analytical techniques like X-ray crystallography and Nuclear Magnetic Resonance (NMR) have been used to determine the structure of RNA. These approaches, though successful, are laborious, costly, and frequently constrained by the dimensions and intricate nature of RNA molecules.
However, AI-enabled algorithms can sort through the available data, identify patterns, and project RNA structures more quickly and precisely. Several studies have demonstrated that machines are now capable of foretelling the three-dimensional structures of RNA molecules, including mRNA—which codes for proteins—and non-coding RNA, which can perform a range of physiological roles. Comparable to proteins, the configurations of single stranded RNA molecules are crucial for their functioning, but little is understood about how these shapes develop and what consequences they have. Although the bases of RNA can form hydrogen-bonded pairs internally, there are a large number of pairing possibilities for RNA molecules with multiple bases. Artificial intelligence is used to identify correlations in the data, and as associations are made, some links relating to particular combinations will be diminished while others will develop.
Contrary to protein folding, which generates an abundance of data that researchers utilize to teach machines, RNA molecule folding does not have this advantage. The structure database known as atomic rotationally equivariant scorer (ARES), in which the structures of 18 tiny RNA molecules were identified experimentally, is an important example of a breakthrough, however. ARES was used to choose the most appropriate structural model for these previously unstudied RNA sequences after receiving the structural models based solely on atomic arrangement and chemical components. Given a sequence and roughly 1,500 potential three-dimensional structures, ARES was able to score each one to determine which one would be most similar to the genuine model. This scoring was done using patterns observed in the training set.
Large genomic datasets can be analyzed by AI algorithms to discover significant players in the metabolism, revealing the molecular machinery underlying RNA stability, turnover, and degradation. Scientists can create tiny compounds or nucleic acids aimed at particular RNA molecules linked to diseases via AI's capacity to predict RNA shapes and interactions. Additionally, AI can speed up the drug development process by helping evaluate large collections of compounds for prospective medications that target RNA molecules.
Reference: Geometric deep learning of RNA structure. Townshend RJL, Eismann S, Watkins AM, Rangan R, Karelina M, Das R, Dror RO. Science. 2021 Aug 27;373(6558):1047-1051.
Luo, Y., Liu, L., He, Z., Zhang, S., Huo, P., Wang, Z., Jiaxin, Q., Zhao, L., Wu, Y., Zhang, D., Bu, D., Chen, R., & Zhao, Y. (2022). TREAT: Therapeutic RNAs exploration inspired by artificial intelligence technology. Computational and structural biotechnology journal, 20, 5680–5689.
Pepe, G., Appierdo, R., Carrino, C., Ballesio, F., Helmer-Citterich, M., & Gherardini, P. F. (2022). Artificial intelligence methods enhance the discovery of RNA interactions. Frontiers in molecular biosciences, 9, 1000205.
Big pharma craves a slice of AI-based RNA drug discovery. (2023). Nature biotechnology, 41(3), 305.
Vadapalli, S., Abdelhalim, H., Zeeshan, S., & Ahmed, Z. (2022). Artificial intelligence and machine learning approaches using gene expression and variant data for personalized medicine. Briefings in bioinformatics, 23(5), bbac191.