基于深度学习和支持向量机的基因结合蛋白预测

Prediction of Gene Binding Protein Based on Deep Learning and Support Vector Machine

针对传统的方法对蛋白质预测的精度低且需要人工提取环节等问题,提出一种基于深度学习和支持向量机的基因结合蛋白预测算法;该算法将卷积神经网络与门控循环单元结合,搜索蛋白质序列,保留蛋白质序列中氨基酸的位置依赖性,利用支持向量机代替神经网络的Softmax分类器对蛋白质的特征序列进行预测;将该模型分别在基准数据集DBP2858和PDB14189上进行对比实验。结果表明,该模型具有更好的脱氧核糖核酸结合蛋白预测能力,并且预测精度和效率均较高。

A iming at the problems of low precision and manual extraction of protein prediction by using traditional methods,a prediction algorithm of gene binding protein based on deep learning and support vector machine was proposed,which combined convolutional neural network with gated recurrent unit.The protein sequences were searched,and position dependence of amino acids in the protein sequences were retained.Support vector machine was used to predict the characteristic sequences of protein instead of Softmax classifier of neural network.The model was tested on benchmark data sets DBP2858 and PDB14189 respectively.The results show that the model has better prediction ability of deoxyribonucleic acid binding protein,and has higher prediction accuracy and efficiency.