Prediction of beta-turn types using SVM and evolutionary information

β-turn, a majority of tight turn which is one of the three most important structural features after α-helix and β-sheet, plays an important role in protein folding and stability. In the past three decades, many methods for predicting β-turns or β-turn types have been developed. In this paper, we perform a novel method of predicting β-turn types using support vector machine, based on evolutionary information generated by PSI-BLAST and secondary structure information produce by PSIPRED. This method is tested on a non-homologous dataset of 426 protein chains. The overall accuracy and MCC of predicting type I, II, IV, VIII and NS (no-specific) β-turn is much better than neural network recently developed by Kaur and Raghava.