一种基于液相色谱-质谱技术进行血清代谢组学研究的方法：从代谢指纹到潜在标志物

本文描述了一种基于液相色谱-质谱技术（LC-MS）的代谢组学发现疾病潜在标志物的方法.该方法利用LC-MS获得代谢指纹图谱,并通过多种统计分析方法对产生的海量数据进行分析,最终筛选出潜在标志物.数据分析过程包括：通过归一化、修正80%规则、数据集分割和数据缩放等方法对数据集进行预处理 通过正交校正的偏最小二乘（OPLS）模式识别方法对样品进行分型 根据模型的变量重要性因子（VIP值）、非参数检验结果和z值筛选潜在标志物.以宫颈癌血清样本为例,应用上述方法,15个变量被确认为潜在标志物,操作者接受曲线（ROC）下的面积为0.667～0.956.经过相关性分析和结构鉴定,发现这15个变量来自9个化合物.其中7个化合物被鉴定为色氨酸、硬脂酸、花生四烯酸、溶血磷脂酰胆碱（0：0/16：0,16：0/0：0,18：1/0：0和18：0/0：0）,说明在宫颈癌中花生四烯酸和溶血磷脂酰胆碱的代谢发生异常.

A metabonomics approach based on liquid chromatography-mass spectrometry： from metabolic profiling to potential biomarker

A LC-MS based metabonomics approach for potential biomarkers screening is described. Metabolic profiles were acquired by LC-MS technology, and potential biomarkers were filtered by multiple statistical methods. Data pretreatment includes data normalization and scaling, corrected 80% rule, and dataset division. According to metabolic patterns, cervical cancer and health group were separated by OPLS. Potential biomarker screening was performed according to VIP value, significant test and z-score. As a result, fifteen variables were considered as potential biomarker, and their AUC were 0.667-0.956. The fifteen variables correspond to nine compounds, seven of which were identified as tryptophan, stearic acid, arachidonic acid, and lysoPC （0：0/16：0, 16：0/0：0, 18： 1/0：0, 8 ： 0/0 ： 0）. It can be concluded that abnormal metabolism of arachidonic acid and lysoPC happened in cervical cancer.