DeCban: Prediction of circRNA-RBP Interaction Sites by Using Double Embeddings and Cross-Branch Attention Networks.

Liangliang Yuan,Yang Yang

doi:10.3389/fgene.2020.632861

Abstract

Circular RNAs (circRNAs), as a rising star in the RNA world, play important roles in various biological processes. Understanding the interactions between circRNAs and RNA binding proteins (RBPs) can help reveal the functions of circRNAs. For the past decade, the emergence of high-throughput experimental data, like CLIP-Seq, has made the computational identification of RNA-protein interactions (RPIs) possible based on machine learning methods. However, as the underlying mechanisms of RPIs have not been fully understood yet and the information sources of circRNAs are limited, the computational tools for predicting circRNA-RBP interactions have been very few. In this study, we propose a deep learning method to identify circRNA-RBP interactions, called DeCban, which is featured by hybrid double embeddings for representing RNA sequences and a cross-branch attention neural network for classification. To capture more information from RNA sequences, the double embeddings include pre-trained embedding vectors for both RNA segments and their converted amino acids. Meanwhile, the cross-branch attention network aims to address the learning of very long sequences by integrating features of different scales and focusing on important information. The experimental results on 37 benchmark datasets show that both double embeddings and the cross-branch attention model contribute to the improvement of performance. DeCban outperforms the mainstream deep learning-based methods on not only prediction accuracy but also computational efficiency. The data sets and source code of this study are freely available at: https://github.com/AaronYll/DECban.

Highlights

Circular RNAs are a special kind of non-coding RNA molecules
To better mine the sequence information, we propose a double-embedding method to expand the feature space, which is further learned by deep neural networks to extract abstract features for classification
Attention mechanism becomes a necessary part to enable the model focus on informative regions, the BiLSTM model with attention improves the performance of long short-term memory networks (LSTMs) significantly, even better than basic convolutional neural networks (CNNs) and ResNets

Summary

Introduction

Different from linear RNAs, circRNA molecules have closed-ring structures, which are not affected by RNA exonuclease, and their expression is more stable (Pamudurti et al, 2017; Li et al, 2018). Emerging studies have shown that circRNAs can bind to various types of proteins to affect protein localization, regulate protein expression, or influence protein-protein-interactions. High-throughput experimental technologies have been widely used to detect the interactions between RNAs and proteins, like cross-linking and immunoprecipitation followed by RNA sequencing (CLIP-Seq) (Yang et al, 2015). The large-scale experimental data makes it possible to predict RNA-protein interactions (RPIs) based on machine learning methods (Li et al, 2013). Compared with expensive and time-consuming wet experiments, the computational methods have considerably sped up the identification of interactions, the automatic prediction of RPI has been a hot topic in the bioinformatics field (Pan et al, 2019)

Objectives

Methods

Results

Discussion

Conclusion