Prediction of Protein B-Factor Profile Based on Feature Selection and Kernel Learning

B-factor reflects the atom´s uncertainty about its average position within a crystal structure and is highly correlated with protein functions. In this article, we propose a novel approach to predict the real value of B-factor. We firstly extract features from the protein sequences and their evolution information, then apply random forest tree to select the important features, which are further inputted to a two-stage support vector regression (SVR) for prediction. Our results have revealed that a systematic analysis of the importance of different features makes us have deep insights into the different contributions of features and is very necessary for developing effective B-factor prediction tools. We thus develop an online Web server, which is freely available at http://www.csbio.situ.edu.cn/bioinf/PredBF for academic use.