PlantC2U: Deep learning of cross-species sequence landscapes predicts plastid C-to-U RNA editing in plants

In plants, C-to-U RNA editing is mainly occurred in the plastids and mitochondria transcripts, which contributes to complex transcriptional regulatory network. More evidences reveal that RNA editing plays critical roles in plant growth and development. However, RNA editing sites accurately detected by transcriptome sequencing data alone are still challenging. In the present study, we developed PlantC2U, which is a convolutional neural network to predict plastid C-to-U RNA editing based on the genomic sequence. PlantC2U achieves over 95% sensitivity and 99% specificity, which outperforms random forest and support vector machine. PlantC2U not only further checks RNA editing sites from transcriptome data to reduce the possible false positives, but also assesses the effect of different mutations on C-to-U RNA editing status based on the flanking sequences. Moreover, we found the patterns of tissue-specific RNA editing in mangrove plant Kandelia obovata, and observed reduced C-to-U RNA editing rates in cold stress response of K. obovata, suggesting their potential regulatory roles in the plants stress adaption. In addition, we present RNAeditDB, available online at https://jasonxu.shinyapps.io/RNAeditDB/. Together, PlantC2U and RNAeditDB would help researchers explore the RNA editing events in plants and thus would be of broad utility for the plant research community. HighlightWe develop a convolutional neural network based deep learning, PlantC2U program, which help researchers explore the plastids C-to-U RNA editing events in plants and thus would be of broad utility for the plant research community.