Improving Protein Expression Prediction Using Extra Features and Ensemble Averaging