“…Some of these are based on the primary sequence of a protein (4,7,9,10,12,15–18,21), whereas others are built using structure-based information (1,2,5,6,8,11,13,14,19,20,22). Machine-learning methods such as support vector machine (SVM) classifiers (15,19), neural networks (1,13) and random forest-based approaches (16,18) have been used for training feature-based models to identify DNA-binding sites.…”