Prediction of FAD Binding Residues with Combined Features from Primary Sequence

In order to analyze the impact of different characteristics of protein sequence on small nucleotide binding site prediction, this paper proposed four sequence-based methods for identifying FAD binding residues of flavin binding proteins (FBP) by means of support vector machine (SVM). We used the different combined features obtained from primary sequence as input for the prediction: evolutionary conservation, predicated solvent accessibility, physicochemical characteristics, residue neighbor list and so on. Our result shows that, the three methods which combined the evolutionary information performed much better than the predictor which did not adopt the evolutionary information. The predictor which combined the feature of residue neighbor list, evolutionary information and physicochemical properties performs the best, achieved accuracy of 87.7326% which is 4.87% higher than the previous methods.