预测同源寡聚蛋白质的支持向量机方法

计算实验表明蛋白质一级结构包含着四级结构信息。本文用支持向量机方法从蛋白质一级结构出发区分同源二聚体和非同源二聚体。蛋白质原始序列的子序列分布用于支持向量机的输入向量，从而充分考虑了蛋白质序列的信息。当子序列的长度为3时，10次交叉验证的总预测准确率达到84．9％，在相同的数据集上，比原有的决策树方法提高了15．0％。实验表明残基顺序对同源寡聚蛋白质的识别起重要作用，而支持向量机方法是蛋白质四级结构预测的强有力工具。

Support vector machine approach for prediction of homo-oligomeric proteins

Computational experiments have shown that protein primary sequence contains quaternary structure information.In this pres-ent work,support vector machine approach is applied to discriminating between homodimers and non-homodimers from the primarystructure.For training and testing protein primary sequences,their subsequence distributions act as input vectors of support vector ma-chine,therefore,the information of protein sequences is sufficiently taken into account.When the length of subsequence is 3,the over-all accuracy of 10-fold cross-validation test is as high as 84. 9%,which increases by 15.0% compared with that of the previous deci-sion tree method on the same data set.Our tests demonstrate that the residue order along protein sequences plays an important role onrecognition of the homo-oligomers and the support vector machine method is a powerful tool for prediction of protein quaternary struc-ture.