Radial basis function method for prediction of protein secondary structure

The paper proposed a new method based on radial basis function neural networks for prediction of protein secondary structure. To make the algorithm comparable to other secondary structure prediction methods, we used the benchmark evaluation data set of 126 protein chains in this paper. We also analyzed how to use evolutionary information to enhance the prediction accuracy. The paper discussed the influence of data selection and structure design on the performance of the networks. The results indicate that this method is feasible and effective.