ANGLOR is a machine-learning based algorithm for ab initio prediction of protein backbone torsion angles. For a given amino acid sequence, the real-value backbone torsion angles (phi and psi) for each residue are predicted by the combination of the neural network training and the support vector machine. References: S. Wu, Y. Zhang. ANGLOR: A Composite Machine-Learning Algorithm for Protein Backbone Torsion Angle Prediction. PLoS ONE 2008; 3: e3400. Registration not required.