DeepOMe: A Web Server for the Prediction of 2'-O-Me Sites Based on the Hybrid CNN and BLSTM Architecture.

Hongyu Li,Yubin Xie,Li Chen,Jian Ren,Huiqin Li,Xiaotong Luo,Zaoli Huang

doi:10.3389/fcell.2021.686894

Abstract

2′-O-methylations (2′-O-Me or Nm) are one of the most important layers of regulatory control over gene expression. With increasing attentions focused on the characteristics, mechanisms and influences of 2′-O-Me, a revolutionary technique termed Nm-seq were established, allowing the identification of precise 2′-O-Me sites in RNA sequences with high sensitivity. However, as the costs and complexities involved with this new method, the large-scale detection and in-depth study of 2′-O-Me is still largely limited. Therefore, the development of a novel computational method to identify 2′-O-Me sites with adequate reliability is urgently needed at the current stage. To address the above issue, we proposed a hybrid deep-learning algorithm named DeepOMe that combined Convolutional Neural Networks (CNN) and Bidirectional Long Short-term Memory (BLSTM) to accurately predict 2′-O-Me sites in human transcriptome. Validating under 4-, 6-, 8-, and 10-fold cross-validation, we confirmed that our proposed model achieved a high performance (AUC close to 0.998 and AUPR close to 0.880). When testing in the independent data set, DeepOMe was substantially superior to NmSEER V2.0. To facilitate the usage of DeepOMe, a user-friendly web-server was constructed, which can be freely accessed at http://deepome.renlab.org.

Highlights

To date, hundreds of different RNA modifications have been identified in human transcriptome, and found to be critical in the regulation of various transcriptional events (Behm-Ansmant et al, 2011)
We present DeepOMe, a web server based on a hybrid deep learning architecture for predicting 2 -O-methylation (2 -O-Me) sites in Human messenger RNAs (mRNAs)
According to the evaluation results, when the flanking sequence length equals to 120 nt and block length equals to 290 nt, the performance generated by our proposed model was the best (AUC = 0.9975, area under Precision-Recall Curve (AUPR) = 0.8818)

Summary

INTRODUCTION

Hundreds of different RNA modifications have been identified in human transcriptome, and found to be critical in the regulation of various transcriptional events (Behm-Ansmant et al, 2011). The previous mentioned high-throughput techniques can provide single-nucleotide mapping of 2 -OMe sites at transcriptome level, the experimental procedure is still expensive and labor-exhausting. Previous studies have randomly sampled subsequences near experimentally identified 2 -O-Me sites as negative sequences. This procedure can produce a high degree of similarity between extracted positive and negative sequences in training dataset, which limits the accuracy of traditional sequence-based models. The development of a reliable prediction tool that can extract useful features from the primary mRNA sequences and produce high-precision results is still an important problem to be solved. We present DeepOMe, a web server based on a hybrid deep learning architecture for predicting 2 -O-Me sites in Human mRNA. A webserver was further developed and makes it easier for researchers and experimenters to use our proposed model

MATERIALS AND METHODS

RESULTS

Experimental Setup

Evaluation Metrics

DISCUSSION