基于LSTM-CNN模型的RNA碱基不成对概率预测研究

Research on the Prediction of RNA Base Mispairing Probability Based on LSTMCNN Model

预测核糖核酸(Ribonucleic Acid,RNA)结构是生物信息领域的热门问题。笔者提出一种长短期记忆(Long Short-Term Memory,LSTM)和卷积神经网络(Convolutional Neural Networks,CNN)相结合的LSTM-CNN深度神经网络模型。该模型基于RNA的一级序列和由LinearFold算法算出的能量最低的二级结构来预测RNA中碱基的不成对概率。最后,使用RNA数据集进行实验。实验结果表明,相对于LSTM模型和CNN模型,LSTM-CNN混合模型有较好的预测效果。

Predicting the Structure of Ribonucleic Acid(RNA) is a hot topic in the field of bioinformatics.In this paper,a deep Neural network model of LSTM-CNN is proposed,which combines Long Short-Term Memory(LSTM) and Convolutional Neural Networks(CNN).The model predicts the probability of base mispairing in the RNA secondary structure based on the primary sequence of RNA and the lowest energy secondary structure calculated by the LinearFold algorithm.Finally,RNA data sets are used for experiments.The experimental results show that LSTM-CNN has a better prediction effect.