摘要
环状RNA(circRNA)是一类新的内源性非编码RNA,广泛存在于真核转录组中,它的环状结构是在剪接过程中通过5'端和3'端共价键反向连接形成.在过去的20多年里,环状RNA可以作为miRNA海绵发挥作用并且在基因调控中,环状RNA与癌症也有密切关系.所以,检测环状RNA对于理解它们的生物作用和生物起源是非常重要的.目前检测环状RNA的一般方法是依赖高通量测序(RNA-Seq),通过在数据中检测RNA的反向剪接位点.然而,由于数据本身和识别的准确度不高,导致其结果假阳性和假阴性偏高.因此,找到一种更准确、更快速的方法识别环状RNA是非常有必要的.为此,提出了基于深度学习的环状RNA识别方法,使用非对称卷积神经网络(ACNN)加双向长短时记忆网络(Bi-LSTM)的架构,利用环状RNA自身的序列特征在长非编码(lncRNA)中进行识别.实验结果表明:所提出的ACNN-BLSTM模型,在各方面性能指标和识别准确率上,都为5种模型中最优,并且识别准确率也达到了90%以上,对比另外四种常见的单一神经网络模型,该方法具有一定的优势.
Circular RNA(circRNA)is a new type of endogenous non-coding RNA,which exists widely in eukaryotic transcriptome and circular structure is formed by the reverse linkage of 5'-end and 3'-end covalent bonds during splicing process.Over the past two decades,circRNA have been shown to function as miRNA sponges and have been closely associated with cancer in gene regulation.Therefore,detection of circRNA is very important for understanding their biological role and biogenesis.At present,the general method for detecting circRNA relies on high-throughput sequencing(RNA-seq),which detects the reverse splicing sites of RNA in the data.However,due to the low accuracy of recognition and data itself,the results of circRNA are high in false positives and false negatives.Therefore,it is necessary to find a more accurate and faster method to identify circRNA.We use the architecture of asymmetric convolutional neural network(ACNN)and bidirectional short and long time memory network(Bi-LSTM)to recognize circRNA in long non-coding(lncRNA)by using the sequence characteristics of circRNA itself.The experimental results show that the ACNN-BLSTM model proposed is the best among the five models in terms of all aspects of performance indicators and recognition accuracy,and the recognition accuracy reaches more than 90%.Compared with the other four common single neural network models,this method has certain advantages.
作者
程威
王帅
范锦江
彭景
林显光
陈恒玲
CHENG Wei;WANG Shuai;FAN Jinjiang;PENG Jing;LIN Xianguang;CHEN Hengling(Biomedical Engineering College,South-Central Minzu University,Wuhan 430074,China)
出处
《中南民族大学学报（自然科学版）》
CAS
北大核心
2022年第6期697-705,共9页
Journal of South-Central University for Nationalities：Natural Science Edition
基金
国家自然科学基金资助项目(31870771)。
关键词
环状RNA
非对称卷积神经网络
双向长短时记忆网络
RNA识别
circular RNA
asymmetric convolutional neural networks
bidirectional long short-term memory network
RNA recognition
