Prediction of DNA-binding residues in proteins from amino acid sequences using a random forest model with a hybrid feature.