Network-Based Analysis and Characterization of Adverse Drug-Drug Interactions
Network-Based Analysis and Characterization of Adverse Drug-Drug Interactions
复制标题
DOI：
10.1021/ci200367w
复制
发表时间：
2011-11-01
影响因子：
5.6
通讯作者：
Kanehisa, Minoru
中科院分区：
文献类型：
--
作者：
Takarabe, Masataka;Shigemizu, Daichi;Kanehisa, Minoru
多种药物联合给药可能会引起不良反应，这些不良反应通常是已知的，但有时是未知的。处方药的说明书应该包含禁忌症和关于不良相互作用的警告，但这些信息不一定完整。因此，为卫生专业人员提供有关所有使用中药物之间药物相互作用的全面观点以及识别潜在相互作用的计算方法变得更加重要，这也可能具有社会实用价值。在这里，我们从日本上市的处方药的所有包装说明书中提取了1，306，565种已知的药物相互作用。它们减少到45，180种相互作用，涉及KEGG药物数据库中D编号识别的1352种药物（活性成分），其中涉及735种药物的14，441种相互作用与相同的药物代谢酶和/或重叠的药物靶点有关。与重叠靶标的相互作用进一步分为三种类型：作用于同一靶标，作用于同一蛋白质家族中不同但相似的靶标，以及作用于属于同一途径的不同靶标。对于提取的其余相互作用数据，我们试图根据解剖学治疗化学（ATC）分类系统定义的药物组来表征相互作用模式，其中D数字水平的高分辨率网络逐渐减少到低分辨率全球网络。基于这项研究，我们已经开发了一个药物相互作用检索系统在KEGG药物数据库，这可能是用于搜索对已知的药物相互作用和预测潜在的相互作用。
Co-administration of multiple drugs may cause adverse effects, which are usually known but sometimes unknown. Package inserts of prescription drugs are supposed to contain contraindications and warnings on adverse interactions, but such information is not necessarily complete. Therefore, it is becoming more important to provide health professionals with a comprehensive view on drug-drug interactions among all the drugs in use as well as a computational method to identify potential interactions, which may also be of practical value in society. Here we extracted 1,306,565 known drug-drug interactions from all the package inserts of prescription drugs marketed in Japan. They were reduced to 45,180 interactions involving 1352 drugs (active ingredients) identified by the D numbers in the KEGG DRUG database, of which 14,441 interactions involving 735 drugs were linked to the same drug-metabolizing enzymes and/or overlapping drug targets. The interactions with overlapping targets were further classified into three types; acting on the same target, acting on different but similar targets in the same protein family, and acting on different targets belonging to the same pathway. For the rest of the extracted interaction data, we attempted to characterize interaction patterns in terms of the drug groups defined by the Anatomical Therapeutic Chemical (ATC) classification system, where the high-resolution network at the D number level is progressively reduced to a low-resolution global network. Based on this study we have developed a drug-drug interaction retrieval system in the KEGG DRUG database, which may be used for both searching against known drug-drug interactions and predicting potential interactions.