基于氨基酸关联的蛋白质—ATP配体结合残基的预测

Prediction of protein-ATP ligand binding residuces based on amino acid association

腺嘌呤核苷三磷酸(ATP)作为一种小分子配体与蛋白质结合对生物体内各种生命活动有重要影响,因此,准确预测蛋白质与ATP配体结合残基具有重要意义。我们在片段水平上对氨基酸的紧邻关联、9个残基对和中心模块进行了统计分析并给出了三种新的特征,分别添加这三种特征及其组合后应用支持向量机(SVM)算法预测蛋白质—ATP配体结合残基,预测结果都有提升,同时添加三种新特征时预测结果最好,5交叉检验下Sn、Sp、Acc值分别达到了79.67%、79.75%、79.71%,MCC和AUC值分别达到0.594和0.879,独立检验的Sn、Sp、Acc、MCC和AUC值分别为80.89%、79.7%、79.77%、0.254和0.887。

As a kind of small molecule ligand,the binding of Adenosine TriPhosphate(ATP)to proteins has an important effect on various life activities in living organisms.Therefore,it is of great significance to accurately predict the binding residue of proteins with ATP ligands.The amino acid adjacency associations,9 residue pairs and central modules were statistically analyzed at the fragment level,and three new features were presented.After respectively adding these three features and their combinations,the support vector machine(SVM)algorithm was applied to predict protein-ATP ligand binding residues,and the prediction results were improved.The best prediction results were obtained when the three features were added at the same time.The values of Sn,Sp,Acc reached 79.67%,79.75%,79.71%respectively and the values of MCC and AUC reached 0.594 and 0.879 respectively under the 5-fold cross-test,and the values of Sn,Sp,Acc,MCC and AUC were 80.89%,79.7%,79.77%,0.254,and 0.887 respectively under the independent test.