H9, HCT116, Xenopus, Mouse Transcriptome

Inosine (I) is a prevalent RNA modification in animals and is formed when an adenosine (A) is deaminated by the ADAR family of enzymes. Traditionally, inosines are identified indirectly as variants from Illumina RNA sequencing (RNA-seq) data because they are interpreted as guanosines by cellular machineries. However, this indirect method performs poorly in protein-coding regions where exons are typically short, in non-model organisms with sparsely annotated single nucleotide polymorphisms (SNPs), or in disease contexts like cancer where unknown DNA mutations are pervasive. Here, we show that Oxford Nanopore direct RNA-seq can be used to identify inosine-containing sites in native transcriptomes with high accuracy. We trained convolutional neural network models to distinguish inosine from adenosine and guanosine, and to estimate the modification rate at each editing site. Furthermore, we demonstrated their utility on the transcriptomes of human, mouse, and Xenopus. Our approach expands the toolkit for the study of A-to-I RNA editing and can be extended to investigate other RNA modifications.

肌苷(Inosine, I)是动物体内广泛存在的RNA修饰类型，由腺苷(adenosine, A)经ADAR酶家族催化脱氨生成。传统研究中，肌苷主要通过Illumina RNA测序（RNA-seq）数据间接鉴定为变异位点，原因是细胞内的分子机制会将肌苷误识别为鸟苷(guanosine)。但该间接鉴定方法存在诸多局限：在外显子普遍较短的蛋白质编码区、单核苷酸多态性(single nucleotide polymorphisms, SNPs)注释稀疏的非模式生物，或是遍布未知DNA突变的癌症等疾病背景中，其识别精度与可靠性均较差。本研究证实，牛津纳米孔直接RNA测序（Oxford Nanopore direct RNA-seq）可用于高精准度鉴定天然转录组中的肌苷修饰位点。我们训练了卷积神经网络(convolutional neural network)模型，以区分肌苷、腺苷与鸟苷，并估算每个编辑位点的修饰效率。此外，我们在人类、小鼠及爪蟾(Xenopus)的转录组中验证了该模型的应用价值。本研究方法拓展了A-to-I RNA编辑的研究工具体系，同时可被推广应用于其他RNA修饰的相关研究。