Improved Prediction Method of Protein Contact Based on RBF Neural Network

In this paper, a prediction method of protein contact on the basis of information granules and RBF neural network have been brought forward. This method improved the encoding approach of protein structure data and classifier performance to enhance the predicting accuracy of protein contact. 200 nonhomologous proteins from the PDB database were encoded according to the encoding approach and were taken as models of training samples. Then the training samples were taken on the granule division based on the information granularity theory and corresponding predicting models for each granule division were set by using RBF neural network. The simulation experiment result indicated that this method can enhance the accuracy of prediction of protein structure.