Multi-Level Association Rule Mining and Network Pharmacology to Identify the Polypharmacological Effects of Herbal Materials and Compounds from Traditional Medicine

An integrated analysis using association rule mining and network pharmacology to identify therapeutic combinations of herbal materials and compounds in traditional medicine. Traditional medicine (TM) has been used to treat a variety of symptoms and diseases through the combination of herbal materials, and it also contributes to the pharmaceutical industry with several advantages such as fewer side effects and significant cost reductions. However, the rules for combining ingredients are not well organized, and complex multi-compound characteristics make it difficult to understand the pharmacological mechanisms among the herbal materials used in TM. In silico approaches that have been proposed to analyze TM and herbal materials require large amount of high-quality structural information or physicochemical properties or have limitations due to ease of interpretation or scope of analysis. In this work, we proposed an approach named InPETM, that integrates association rule mining (ARM) and network pharmacology analyses to identify polypharamcological effects of herbal materials and compounds from TM. Specifically, InPETM performs analyses combining ARM and network pharmacology-based method at the herb-level and compound-level, respectively, and identifies potential herbal material combination and compound candidates for the phenotype. InPETM provided results of pharmacological effects of herbal material combination and compound and identification of mechanism of action in human protein interactome network, which were confirmed by further structural network analysis and literature review analysis. These results indicate that InPETM can contribute to drug development in TM through better understanding of polypharmacological features of herbal materials.本研究采用关联规则挖掘（association rule mining）与网络药理学（network pharmacology）相结合的整合分析方法，旨在识别传统医学（Traditional Medicine，TM）中草药原料与化合物的治疗性组合。 传统医学（Traditional Medicine，TM）通过草药原料的组合用于治疗多种症状与疾病，同时其具备副作用更少、成本显著降低等优势，可为制药产业发展提供支撑。然而，当前草药原料的配伍规则尚未得到系统梳理，且其复杂的多化合物特性使得人们难以理解传统医学中所用草药原料间的药理机制。现有用于分析传统医学与草药原料的计算机模拟（in silico）方法，要么需要大量高质量的结构信息或理化性质数据，要么在解释性或分析范围上存在局限性。 本研究提出了一种名为InPETM的分析方法，该方法整合关联规则挖掘与网络药理学分析，以识别传统医学中草药原料与化合物的多药理效应（polypharmacological）。具体而言，InPETM分别在草药层级与化合物层级开展关联规则挖掘（ARM）与基于网络药理学的分析，并针对目标表型识别潜在的草药原料组合与化合物候选物。InPETM可输出草药原料组合与化合物的药理效应结果，并在人类蛋白质相互作用组网络（human protein interactome network）中识别其作用机制，上述结果通过进一步的结构网络分析与文献综述分析得到了验证。上述结果表明，InPETM可通过加深对草药原料多药理特性的理解，助力传统医学领域的药物研发。