基于集成支持向量机的蛋白质K丙二酰化位点的预测

Prediction of Protein K Malonylation Site Based on Ensemble Support Vector Machine

赖氨酸K丙二酰化是近些年来新发现的一种重要的蛋白质翻译后修饰(PTMs),对生物生长起着至关重要的作用。目前,通过蛋白质组的方法,部分丙二酰化修饰位点已经能够在实验中获取到,然而实验上耗费的时间和仪器上所花费的成本依然面临挑战。因此,开发一种能够准确预测丙二酰化位点的计算方法是有必要的。该系统的工作原理是:⑴将这些蛋白质组合成伪氨基酸(PseAAC);⑵使用下采样来平衡训练集数据集;⑶通过集成支持向量机建立一个综合的预测系统,最终,该系统独立测试集的准确率达到76.46%,效果比较明显,为生物信息学的研究提供帮助。

Lysine malonylation is an important protein post translational modification(PTM)newly discovered in recent years,which plays a vital role in biological growth.At present,in experiments some of the malonylation sites can be obtained through the method of proteomics,but the time spent on experiments and the cost of instruments still face challenges.Therefore,it is necessary to develop a calculation method that can accurately predict malonylation sites.The working principles of the system are:(1)combining these proteins into pseudo amino acids(PseAAC),(2)using down-sampling to balance the training set data set,and(3)building a comprehensive prediction system through ensemble support vector machines.The accuracy rate of the independent test set reaches 76.46%,whose effect is so obvious that it can provide help for the discovery of bioinformatics.