Wang Yin-Ying, Nacher Jose C, Zhao Xing-Ming
Department of Mathematics, Shanghai University, Shanghai 200444, China.
Mol Biosyst. 2012 Apr;8(5):1528-34. doi: 10.1039/c2mb05450g. Epub 2012 Mar 8.
The identification of interactions between drugs and proteins plays key roles in understanding mechanisms underlying drug actions and can lead to new drug design strategies. Here, we present a novel statistical approach, namely PDTD (Predicting Drug Targets with Domains), to predict potential target proteins of new drugs based on derived interactions between drugs and protein domains. The known target proteins of those drugs that have similar therapeutic effects allow us to infer interactions between drugs and protein domains which in turn leads to identification of potential drug-protein interactions. Benchmarking with known drug-protein interactions shows that our proposed methodology outperforms previous methods that exploit either protein sequences or compound structures to predict drug targets, which demonstrates the predictive power of our proposed PDTD method.
药物与蛋白质之间相互作用的识别在理解药物作用机制方面起着关键作用，并可引领新的药物设计策略。在此，我们提出一种新颖的统计方法，即基于药物与蛋白质结构域之间的衍生相互作用来预测新药潜在靶蛋白的PDTD（利用结构域预测药物靶点）方法。那些具有相似治疗效果的药物的已知靶蛋白使我们能够推断药物与蛋白质结构域之间的相互作用，进而识别潜在的药物 - 蛋白质相互作用。与已知药物 - 蛋白质相互作用进行基准测试表明，我们提出的方法优于以往利用蛋白质序列或化合物结构来预测药物靶点的方法，这证明了我们提出的PDTD方法的预测能力。
