Protein Secondary Structure Prediction Based on Generative Confrontation and Convolutional Neural Network

In the field of bioinformatics, the prediction of protein secondary structure is a challenging task, and it is extremely important for determining the structure and function of proteins. In this paper, the generation of adversarial network and convolutional neural network model are combined for protein secondary structure prediction. Firstly, generate a confrontation network to extract protein features, and then combine the extracted features with the original PSSM data as the input of the convolutional neural network to obtain prediction results. Testsets CASP9, CASP10, CASP11, CASP12, CB513 and PDB25 obtained 87.06%, 87.24%, 87.31%, 87.39%, 88.13% and 88.93%, which are 3.88%, 4.6%, 7.97%,5.85%, 5.78%, 4.25% higher than one using the convolutional neural network alone. The experimental results show that the feature extraction ability of generating adversarial networks is very significant.