基于长度信息和深度卷积神经网络分类建模的蛋白质二级结构预测方法

PROTEIN SECONDARY STRUCTURE PREDICTION METHOD BASED ON LENGTH INFORMATION AND DEEP CONVOLUTIONAL NEURAL NETWORK CLASSIFICATION MODELING

提出基于蛋白质长度信息和深度卷积神经网络分类建模的方法(Length Information and Deep Convolutional Neural Networks, LIM-DCNN),实现对于蛋白质二级结构的预测。实验得到的6分段模型,预测CASP9、CASP10、CASP11、CASP12和CB513的Q3准确率分别为83.67%、78.99%、78.53%、71.52%和85.94%,说明了基于蛋白质长度分类建模的有效性,并且实验得到的CB513结果明显优于其他许多经典的预测算法。

This paper proposes a method based on protein length information and deep convolutional neural network(LIM-DCNN) classification modeling to predict the secondary structure of proteins. The 6-segment model obtained by the experiments predicts Q3 accuracy rates of CASP9, CASP10, CASP11, CASP12, and CB513 of 83.67%, 78.99%, 78.53%, 71.52%, and 85.94% respectively. The experiment proves the validity of classification based on protein length, and the CB513 result obtained by the experiment is obviously better than many other classic prediction algorithms.