Department of Computer Engineering, Middle East Technical University, Ankara, Turkey.
Department of Computer Engineering, TOBB University, Ankara, Turkey.
BMC Bioinformatics. 2018 Apr 12;19(1):136. doi: 10.1186/s12859-018-2142-1.
Drug repositioning is the process of identifying new targets for known drugs. It can be used to overcome problems associated with traditional drug discovery by adapting existing drugs to treat new discovered diseases. Thus, it may reduce associated risk, cost and time required to identify and verify new drugs. Nowadays, drug repositioning has received more attention from industry and academia. To tackle this problem, researchers have applied many different computational methods and have used various features of drugs and diseases.
In this study, we contribute to the ongoing research efforts by combining multiple features, namely chemical structures, protein interactions and side-effects to predict new indications of target drugs. To achieve our target, we realize drug repositioning as a recommendation process and this leads to a new perspective in tackling the problem. The utilized recommendation method is based on Pareto dominance and collaborative filtering. It can also integrate multiple data-sources and multiple features. For the computation part, we applied several settings and we compared their performance. Evaluation results show that the proposed method can achieve more concentrated predictions with high precision, where nearly half of the predictions are true.
Compared to other state of the art methods described in the literature, the proposed method is better at making right predictions by having higher precision. The reported results demonstrate the applicability and effectiveness of recommendation methods for drug repositioning.
药物重定位是指为已知药物确定新靶点的过程。它可以通过将现有药物改编为治疗新发现的疾病来克服传统药物发现所带来的问题。因此，它可以降低识别和验证新药的相关风险、成本和时间。如今，药物重定位受到了业界和学术界的更多关注。为了解决这个问题，研究人员已经应用了许多不同的计算方法，并使用了药物和疾病的各种特征。
在这项研究中，我们通过结合多种特征，即化学结构、蛋白质相互作用和副作用，来预测目标药物的新适应症，为正在进行的研究工作做出了贡献。为了实现我们的目标，我们将药物重定位视为推荐过程，这为解决问题提供了一个新视角。所使用的推荐方法基于帕累托优势和协同过滤。它还可以整合多个数据源和多个特征。在计算部分，我们应用了几种设置并比较了它们的性能。评估结果表明，与文献中描述的其他最先进的方法相比，该方法通过更高的精度实现了更集中的预测，其中近一半的预测是正确的。
与文献中描述的其他最先进的方法相比，该方法通过更高的精度实现了更准确的预测。报告的结果表明，推荐方法在药物重定位方面具有适用性和有效性。
