基于回归模型的网络排序算法预测药物靶点关系

计算方法预测新的药物靶点关系是新药研发的主要途径,基于相似药物较容易与相似靶蛋白绑定的假定和标签传播的框架,提出一种基于脊回归模型的网络排序计算方法挖掘潜在的药物靶点作用关系;并对四个数据集分别进行30次10—CV预测分析。结果表明,与现有网络算法DBSI、TBSI和NBI相比,该预测算法能得到较高的ROC曲线面积,具有较好的应用前景。

A Regression-Based Network Ranking Prediction Model for Drug-Target Interaction

Abstract: Computational prediction of potential interaction between drug and target protein is critical for drug discovery and design. In this paper, based on the hypothesis that similar drugs often target similar target proteins and the framework of label propagation, a method of ridge regression based network rank method is developed to predict drug-target interactions. Extensively testing was applied on four datasets. Through comparison with other recently proposed DBSI, TBSI and NBI methods, our algorithm attains high AUC values and is capable of predicting more reliable DTIs.