RNA language models predict mutations that improve RNA function

• Published in: Nature communications 2024-12, Vol.15 (1), p.10627-17
• Main Authors: Shulgina, Yekaterina, Trinidad, Marena I., Langeberg, Conner J., Nisonoff, Hunter, Chithrananda, Seyone, Skopintsev, Petr, Nissley, Amos J., Patel, Jaymin, Boger, Ron S., Shi, Honglue, Yoon, Peter H., Doherty, Erin E., Pande, Tara, Iyer, Aditya M., Doudna, Jennifer A., Cate, Jamie H. D.
• Format: Article
• Language: English
• Subjects: 631/114/129; 631/114/1305; 631/92/500; Biological activity; Catalysis; Data acquisition; Deep Learning; E coli; Escherichia coli - genetics; Escherichia coli - metabolism; Functional analysis; Garnets; Gene expression; Gene sequencing; Genomic analysis; Humanities and Social Sciences; multidisciplinary; Mutation; Nucleic Acid Conformation; Nucleotide sequence; Phenotypes; Predictions; Ribonucleic acid; Ribosomes - genetics; Ribosomes - metabolism; RNA; RNA - genetics; RNA - metabolism; RNA, Ribosomal - chemistry; RNA, Ribosomal - genetics; rRNA; Science; Science (multidisciplinary); Structure-function relationships; Taxonomy; Thermal stability
• ISSN: 2041-1723
• DOI: 10.1038/s41467-024-54812-y

Structured RNA lies at the heart of many central biological processes, from gene expression to catalysis. RNA structure prediction is not yet possible due to a lack of high-quality reference data associated with organismal phenotypes that could inform RNA function. We present GARNET (Gtdb Acquired RNa with Environmental Temperatures), a new database for RNA structural and functional analysis anchored to the Genome Taxonomy Database (GTDB). GARNET links RNA sequences to experimental and predicted optimal growth temperatures of GTDB reference organisms. Using GARNET, we develop sequence- and structure-aware RNA generative models, with overlapping triplet tokenization providing optimal encoding for a GPT-like model. Leveraging hyperthermophilic RNAs in GARNET and these RNA generative models, we identify mutations in ribosomal RNA that confer increased thermostability to the Escherichia coli ribosome. The GTDB-derived data and deep learning models presented here provide a foundation for understanding the connections between RNA sequence, structure, and function. Generating RNA sequences with improved function remains challenging. Here, authors present an RNA database for RNA structural and functional analysis. They use this database and the RNA generative models to identify RNA mutations that increase the thermostability of a bacterial ribosome.