A novel method of predicting gamma-turns using SVM and multiple alignment profiles

From both the structural and functional point of view, tight turns play an important role as well as α-helix and β-sheet, γ-turn, the second most characterized and commonly found tight turn after the β-turn, is defined as a three-residue turn. Recently, Kaur and Raghava developed a neural network method to predict γ-turns and achieved perfect results. In this paper, we employ a SVM method using both evolutionary information and secondary structure information to predict γ-turns. Compared to the neural network method, our method improves several performance measures, such as prediction accuracy, and MCC. Also we use a new performance measure, SOV, to assess our method.