Abstract: Based on the data from the project “Study on the Proteomics about Phlegm and Blood Stasis Symptom in Hyperlipidemia and Atherosclerosis”, multivariate analysis are used to help find the most powerful protein groups for TCM symptom diagnosis Clustering methods on both variables and samples are used to analyze the 11 proteins, and two possibilities for the variable clustering results are given from the TCM perspective to supervise the sample clustering in different ways Moreover, hypothesis tests are introduced to support the classification independently Either the protein group {Haptoglobin (Precursor),α-trypsin inhibitor light chain, ALB protein3, Complement component C4} or the protein group {Fibrinogen gamma chain, α-trypsin inhibitor light chain, undetermined protein (ID1485)} is found to excellently separate not only the symptoms and the control group, but also the phlegm and the blood stasis symptoms Therefore, suggestions for better TCM symptom diagnosis are given out

Key words: symptoms in Traditional Chinese Medicine, variable clustering, sample clustering, hypothesis test, symptom diagnosis

摘要： 从“高脂血症及动脉粥样硬化痰瘀证候的蛋白质组学研究”的数据出发，研究影响中医证候的各主要因素。对11种可能的标志蛋白质（群）数据进行分析，以统计聚类为主导思想，给出变量聚类和样本数据聚类综合应用的方法，并结合医学角度对变量聚类结果的分析，指导组内和组间两种样本聚类讨论；同时，通过假设检验，从统计理论上对所得分类予以支持。最后得到标志蛋白质群{结合珠蛋白前体，α-胰蛋白酶抑制剂轻链，脂肪细胞脂质结合蛋白异构体3，补体C4}或{纤维蛋白原γ链，α-胰蛋白酶抑制剂轻链，未确定名称的蛋白(ID1485)}。考虑可能是区分高脂血症及动脉粥样硬化痰证和瘀证的标志蛋白质群，从而给出蛋白质水平上对痰证和瘀证判决方法的建议。

关键词: 中医证候, 变量聚类, 样本数据聚类, 假设检验, 判决方法