2026-11-12 東京大学

概要図:ナノポア直接RNAシーケンスと深層学習によるtRNAの統合解析
<関連情報>
細胞内tRNAのナノポアシーケンシングと深層学習分類器による絶対定量およびプロファイリング(Absolute quantification and profiling of cellular tRNAs by nanopore sequencing and a deep learning classifier)
Hiroki Ueda, Mai Maeda, Shunpei Okada, Ryo Noguchi, Bhaskar Dasgupta, Ryo Yamamoto, Miho Furuta, Ayumi Nishi, Asuteka Nagao, Yuriko Sakaguchi, Kenjyo Miyauchi, Tsutomu Suzuki
Nature Communications, 17, Article 11332 (2026)2026-11-09
DOI:10.1038/s41467-026-78563-0
Transfer RNAs (tRNAs) are essential molecules in protein synthesis, delivering amino acids to ribosomes according to mRNA codons. Their abundance and modification states influence translation efficiency and cellular functions. However, the extensive and diverse chemical modifications of tRNAs introduce errors during reverse transcription and sequence basecalling, making accurate quantification challenging with conventional sequencing methods. In this study, researchers developed the nano tRNA classifier, a deep learning framework that directly analyzes ionic current signals from nanopore direct RNA sequencing (DRS) without converting them into nucleotide sequences. Trained using 44 Escherichia coli tRNA species, the classifier achieved an average classification accuracy of 99.1% while retaining 82.7% of sequencing reads, outperforming conventional sequence-mapping approaches. By correcting sequencing biases using an equimolar tRNA reference mixture, the researchers further established absolute quantification of all 48 E. coli tRNA species. They also developed binary classifiers capable of identifying specific RNA modifications at the single-molecule level, including pseudouridine and other modifications. This integrated platform enables accurate profiling of both tRNA abundance and modification states, providing a new analytical foundation for investigating translational regulation, cellular responses to environmental changes, and RNA-related alterations in disease.

