A comparison of two machine learning methods for protein secondary structure prediction

Nowadays, the best methods for protein secondary structure prediction are based on neural network and support vector machine, and both of them incorporate the information from multiple sequences alignment. However the two methods were executed on different training and testing data sets. A comparison between the two methods has been carried on here. We use the most stringent cross validation test procedure to assess the two methods on CB513, which is one of the most popular used data sets. Neural network achieved a Q/sub 3/ accuracy of 74.2%, while support vector machine got Q/sub 3/ of 76.6%, which was slightly better than NN.