目的 整合化学成分、化学特征探究中药属性与化合物之间的关联性。方法 从公共数据平台收集常用中药的化学成分及结构信息，计算成分指纹及其相似性，构建中药属性的化学特征向量，分析不同属性分类的化学特征。通过BM25-adpt算法评价化学特征对中药属性的贡献度，并计算成分种类得分，并选取活血药关键成分进行富集分析。采用Kendall’s W一致性检验评估药性、药味、归经及功效分类成分种类得分一致性。结果 共获得437味中药的44 279个成分，以0.5为相似度阈值筛选出5 016个成分种类。Kendall’s W一致性检验结果显示，4个属性的c_score得分集存在显著差异。余弦相似度分析进一步表明，不同程度药性分类的化学特征存在差异。药味和归经的复合分类中有相同组成的分类在化学特征上存在一定的联系，且药味中苦与微苦相似度较高。针对“活血行气止痛”与“活血调经”2类功效的代表性成分进行基因本体（gene ontology，GO）功能和京都基因与基因组百科全书（Kyoto encyclopedia of genes and genomes，KEGG）通路富集分析，发现其药理作用可能通过神经活性配体-受体相互作用、类固醇激素生物合成过程、调节氧化还原酶活性等过程实现。结论 中药属性与其化学成分之间存在关联，每种属性分类存在关键成分群。不同程度的药性分类不可简单视为一类，药味及归经分类中的复合分类存在一定关联，相似作用趋势的功效分类在化学特征上体现出一定的相似性。研究结果为中药药效物质基础研究及方剂配伍机制的现代化阐释提供了数据支持。

Objective To investigate the association between traditional Chinese medicine (TCM) properties and chemical compounds by integrating chemical composition and chemical characteristics. Methods Chemical constituents and structural data of commonly used TCMs were collected from public databases. Molecular fingerprints and pairwise similarities were computed to construct chemical feature vectors representing TCM properties, enabling comparative analysis across different property categories. The BM25-adpt algorithm was applied to assess the contribution of chemical features to each TCM property, yielding component-type scores. Key components of blood-activating herbs were selected for functional enrichment analysis. Kendall’s W concordance test was used to evaluate the consistency of component-type scores across four property dimensions: medicinal properties, medicinal flavors, meridian tropism, and therapeutic functions. Results A total of 44 279 chemical components from 437 TCMs were compiled. At a structural similarity threshold of 0.5, these were clustered into 5 016 distinct component types. Kendall’s W test revealed significant discrepancies among the c_score distributions across the four property categories. Cosine similarity analysis further demonstrated that different levels of medicinal properties exhibit distinct chemical profiles. Composite classifications within medicinal flavors and meridian tropism that share identical constituent sets showed notable chemical relatedness; in particular, “bitter” and “slightly bitter” flavors exhibited high similarity. Enrichment analyses of gene ontology (GO) functions and Kyoto encyclopedia of genes and genomes (KEGG) pathways were performed on representative components of the two efficacies “blood activating, qi-moving and analgesic” and “blood activating to regulate menstruation” implicated key biological processes including neuroactive ligand-receptor interactions, steroid hormone biosynthesis, and regulation of oxidoreductase activity. Conclusion This study establishes a measurable link between TCM properties and their underlying chemical constituents, with each property category characterized by specific clusters of key components. Different gradations of medicinal properties should not be treated as homogeneous, composite categories within medicinal flavors and meridian tropism show meaningful chemical associations; and therapeutic functions with similar clinical actions demonstrate convergent chemical features. These findings provide a data-driven foundation for elucidating the material basis of TCM efficacy and advancing the modern understanding of herbal compatibility mechanisms.

TP18;R285.1

国家自然科学基金面上项目（82474352）；湖南省中医药管理局重点项目（A2024011）