About 50 years ago, the field of molecular biology began to uncover the mechanisms behind gene function, shedding light on changes in forms, adaptations, complexity, and the basis of human diseases. Researchers have explored various processes such as gene birth, gene duplication, gene expression regulation, and splicing regulation. Now, with the advent of big data and artificial intelligence (AI), scientists are delving into a particularly fascinating and elusive mechanism: RNA editing.
RNA editing is a process where a single nucleotide in an RNA molecule is changed. This seemingly small alteration can significantly increase the complexity of the transcriptome (the set of all RNA molecules) and the proteome (the set of all proteins). In essence, RNA editing allows for more variation and adaptability in how genes are expressed and how proteins are formed.
To better understand RNA editing, researchers at Universitat de Barcelona have developed a new approach that uses AI. They employed two advanced AI algorithms: random forest (RF) and bidirectional long short-term memory (biLSTM) neural networks with an attention layer. These algorithms were trained using data from RNA-editing databases and variant calling (identifying differences) from RNA and DNA sequencing experiments conducted on the same individuals across different species.
The AI algorithms analyzed both the primary sequence (the order of nucleotides) and the secondary structure (the three-dimensional shape) of RNA to predict RNA-editing events. By doing so, the researchers could identify where and how RNA editing occurs.
To further understand RNA editing, the team devised a novel method called cross-testing analysis. This method allows them to assess whether the mechanisms of RNA editing are conserved (remain the same) or diverge (change) across different species, all done entirely through computer simulations (in silico). This approach not only enhances our understanding of how RNA editing has evolved but also provides insights into the specific mechanism of adenosine-targeting, which is a key part of RNA editing.
This innovative research combining AI and molecular biology offers a deeper understanding of RNA editing, highlighting its role in increasing biological complexity. The use of AI to predict and analyze RNA editing events opens new avenues for research, potentially leading to breakthroughs in understanding gene regulation and developing new treatments for diseases.
Zawisza-Álvarez M, Peñuela-Melero J, Vegas E, Reverter F, Garcia-Fernàndez J, Herrera-Úbeda C. (2024) Exploring functional conservation in silico: a new machine learning approach to RNA-editing. Brief Bioinform 25(4):bbae332. [article]