Attention-based multi-label neural networks for integrated prediction and interpretation of twelve widely occurring RNA modifications

Zitao Song,Kunqi Chen,Yiyou Song,Jionglong Su,Bowen Song,Daiyun Huang,João Pedro De Magalhães,Daniel J Rigden,Jia Meng,Gang Liu

doi:10.1038/s41467-021-24313-3

Abstract

Recent studies suggest that epi-transcriptome regulation via post-transcriptional RNA modifications is vital for all RNA types. Precise identification of RNA modification sites is essential for understanding the functions and regulatory mechanisms of RNAs. Here, we present MultiRM, a method for the integrated prediction and interpretation of post-transcriptional RNA modifications from RNA sequences. Built upon an attention-based multi-label deep learning framework, MultiRM not only simultaneously predicts the putative sites of twelve widely occurring transcriptome modifications (m6A, m1A, m5C, m5U, m6Am, m7G, Ψ, I, Am, Cm, Gm, and Um), but also returns the key sequence contents that contribute most to the positive predictions. Importantly, our model revealed a strong association among different types of RNA modifications from the perspective of their associated sequence contexts. Our work provides a solution for detecting multiple RNA modifications, enabling an integrated analysis of these RNA modifications, and gaining a better understanding of sequence-based RNA modification mechanisms.

Highlights

Recent studies suggest that epi-transcriptome regulation via post-transcriptional RNA modifications is vital for all RNA types
MultiRM consists of an embedding module representing the input RNA sequences using the inherent short- and long-range interactions among nucleotides
The embedded representation is fed to an long short-term memory (LSTM) layer to extract the underlying sequence features shared by all modifications

Summary

Introduction

Recent studies suggest that epi-transcriptome regulation via post-transcriptional RNA modifications is vital for all RNA types. Precise identification of RNA modification sites is of crucial importance to understanding the functions and regulatory mechanisms of various RNAs. More than 100 different types of RNA modifications have been identified[2], and among them, N6-methyladenosine (m6A) is the most common eukaryotic mRNA modification. M6A occurs on nascent pre-mRNA, regulating its stability and translation It is involved in many biological processes such as the circadian clock, differentiation from naïve pluripotency, and the heat shock response. Special attention has been paid to the prediction of RNA modifications in introns[23], lncRNAs24 as well as various tissues and cell lines[25,26,27] Together, these works greatly advanced our understanding of the localization of multiple RNA modification types in different species under various conditions[28]. Given a large number of negative (non-modifiable) sites of such rare RNA modifications, the sequence-based prediction is likely to produce a substantial proportion of false-positive predictions in practice and should be used with extra caution

Objectives

Methods

Results

Conclusion