文献类型：期刊文献

中文题名：基于分组贝叶斯排序的药物-靶标关系预测

英文题名：Drug-Target Interaction Prediction Based on Grouped Bayesian Ranking Approach

作者：丁棋梁[1];石泽智[1];李建华[1]

机构：[1]华东理工大学信息科学与工程学院,上海200237

年份：2020

卷号：56

期号：15

起止页码：185

中文期刊名：计算机工程与应用

外文期刊名：Computer Engineering and Applications

收录：CSTPCD;;北大核心:【北大核心2017】；CSCD:【CSCD_E2019_2020】；

基金：国家科技重大专项项目(No.2018ZX09735002)。

语种：中文

中文关键词：药物-靶标相互作用;机器学习;贝叶斯排序;分组策略

外文关键词：drug-target interaction;machine learning;Bayesian ranking;grouping strategy

摘要：基于贝叶斯排序的药物-靶标关系预测已经取得较好效果,但忽略了同一靶标的药物间的关联关系,从而影响精度。针对此问题,提出了一种新方法——基于分组贝叶斯排序的药物-靶标关系预测。在该方法中,根据与特定靶标存在相互作用的药物间具有相似性的现实,引入分组策略使这些相似药物间产生互动,并推导出基于分组策略的理论模型。该方法在五个公开数据集上与五种典型方法进行对比,产生的结果均优于所对比的方法。
Drug-target interaction prediction based on Bayesian ranking has achieved good results,but neglects the correlation between drugs of the same target,which affects the accuracy.To solve this problem,a new method of drug-target interaction prediction based on grouped Bayesian ranking is proposed.In this method,according to the similarity of drugs interacting with specific targets,the grouping strategy is introduced to link these similar drugs,and the theoretical model based on grouping strategy is deduced.This method is comparing with five typical methods on five open datasets,and the experimental results are better than the comparison method.