支持向量回归预测蛋白质残基的B因子  

Prediction of B-factor profile from sequence using support vector regression

作　　者：尹辉[1] 敬闰宇[2] 李益洲[1] 文志宁[1] 李梦龙[1] 

机构地区：[1]四川大学化学学院,四川成都610064 [2]四川大学数学学学院,四川成都610064

出　　处：《计算机与应用化学》2011年第11期1477-1480,共4页Computers and Applied Chemistry

基　　金：国家自然科学基金资助(20775052;20972103)

摘　　要：蛋白质分子的柔性在各种生物进程中如酶催化、蛋白质分子绑定和识别等发挥着重要的作用。研究蛋白质分子柔性有助于更好地理解蛋白质的功能,研究证明从X射线晶体衍射结构而来的B因子能够较好的衡量蛋白质分子的柔性。从蛋白质序列出发,以保守信息、预测的二级结构、预测的相对溶剂可及性和氨基酸4种物理化学性质作为特征向量,利用支持向量回归方法对蛋白质的B因子进行预测。考察了邻近残基的影响,采用滑动窗口方法对所有特征变量进行了处理。通过对位置特异得分矩阵的平滑窗口处理,提高了预测精度。以皮尔逊相关系数为算法评价指标,最终得到独立测试集的皮尔逊相关系数为0.56,表明该方法有效提高预测精度。B-factor from X-ray crystal structure can well measure protein structural flexibility,which plays an important role in different biological processes,such as catalysis,binding and molecular recognition.Understanding the essence of flexibility can be helpful for the further study of the protein function.In this study,we predicted B-factor profile from sequence by using support vector regression（SVR）.Here, amino acid sequence information including position-specific scoring matrix（PSSM） profiles,predicted secondary structure,predicted relative solvent accessibility and physicochemical properties were employed to features.To consider the influence of the neighboring residues of an amino acid,a sliding window is used to encode the features.A smoothing window is used to encode PSSM profiles.Pearson correlation coefficient（CC） was used as the performance measure.CC value was 0.56 for the independent dataset.The result showed that the prediction accuracy can be improved.

关 键 词：B因子 平滑窗口 支持向量回归 

分 类 号：Q51[生物学—生物化学] TP319[自动化与计算机技术—计算机软件与理论]