Weakly supervised learning of RNA modifications from low-resolution epitranscriptome data

• DOI: 10.1093/bioinformatics/btab278
• Publication Date: 2021-04-23T20:20:16Z
• Authors: Daiyun Huang; Bowen Song; Jingjue Wei; Jionglong Su; Frans Coenen; Jia Meng
• Keywords: Identification; Nucleic acid structure

Abstract Motivation Increasing evidence suggests that post-transcriptional ribonucleic acid (RNA) modifications regulate essential biomolecular functions and are related to the pathogenesis of various diseases. Precise identification RNA modification sites is for understanding regulatory mechanisms RNAs. To date, many computational approaches predicting have been developed, most which were based on strong supervision enabled by base-resolution epitranscriptome data. However, high-resolution data may not be available. Results We propose WeakRM, first weakly supervised learning framework from low-resolution datasets, such as those generated acRIP-seq hMeRIP-seq. Evaluations three independent datasets (corresponding different types their respective sequencing technologies) demonstrated effectiveness our approach in WeakRM outperformed state-of-the-art multi-instance methods genomic sequences, WSCNN, was originally designed transcription factor binding site prediction. Additionally, captured motifs consistent with existing knowledge, visualization predicted modification-containing regions unveiled potentials detecting improved resolution. Availability implementation The source code algorithm, along used, freely accessible at: https://github.com/daiyun02211/WeakRM Supplementary information available at Bioinformatics online.