基于惩罚岭型线性判别对乙酰氨基酚肝毒性代谢组学的研究

Studies on Metabolomics of Liver Toxicity of Acetaminophen based on Penalized Ridge-type Linear Discriminant Analysis

目的:研究基于惩罚岭型线性判别分析(PRLDA)对乙酰氨基酚(APAP)为肝毒性模型药物,与传统的主成分分析(PCA)比较建立的代谢组学模型的方法学。方法:应用APAP对大鼠肝毒性造模,利用UPLC-MS检测尿液样本,以总离子流色谱为毒性表达变量,运用PRLDA和PCA分别建立APAP肝毒性代谢组学模型。结果:在PRLDA代谢组学模型中,APAP组明显偏离正常组,可以直观看出正常组与模型组的类间差异,具有较强的分类能力;而PCA的分类能力较差。结论:PRLDA代谢组学模型的建立能准确、灵敏地表达药物毒性,并且该方法优于PCA模型。

Objective： To Compare with the principal component analysis （PCA）, the methodology of establishing metabolomics model by penalized ridge-type linear discriminant analysis （PRLDA） would be studied based on the liver toxicity of the model drug acetaminophen （APAP）. Methods： The rat liver toxicity model was made by using APAP. The rats＇ urine samples collected were detected by using UPLC-MS. With total ion chromatogram as a variable of ex- pressing toxicity, the metabolomics model on liver toxicity of acetaminophen was established by PRLDA and PCA sep- arately. Results： In the PRLDA metabolomics model, APAP group was deviated from the normal group obviously, and there was a large degree of the difference between the normal group and APAP group. There was a strong classification ability for PRLDA metabolomics model and a poor one for PCA one. Conclusion ： The PRLDA metabolomics model could express the drug toxicity accurately and sensitively, and it is better than the PCA model.