基于动态概率主成分分析模型的钙干预敏感标志物代谢组学研究

项目名称： 基于动态概率主成分分析模型的钙干预敏感标志物代谢组学研究

项目编号： No.81502889

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 医药、卫生

项目作者： 张秋菊

作者单位： 哈尔滨医科大学

项目金额： 18万元

中文摘要： 骨密度作为钙干预研究的主要指标在准确性、灵敏性、应用广泛性、放射性和价格费用较高等方面存在问题，这些问题中以灵敏性差最为突出，在干预研究中不能用于早期干预效果评价和敏感人群识别。本研究通过检测钙干预人群不同时点代谢产物，拟合动态概率主成分分析（DPPCA）模型筛选出有意义的敏感生物标志物，并进一步探索代谢途径。主要研究内容包括：动态概率主成分模型的构建和模拟运算、模型参数估计、时间效应估计及标志物显著性检验、数据可视化、标志物鉴定和生物学解释等几个方面。本研究预期构建适合多时点动态代谢组数据的动态概率主成分分析模型，给出模型参数估计方法和策略。该模型能够分别对每个处理组的代谢产物进行时间效应估计，筛选出干预组和对照组时间变化显著的代谢物。该研究可以为钙干预试验中敏感标志物的识别提供有效可靠的信息，进而可以对敏感人群进行筛查，对调整补钙策略、人群健康和疾病预防具有很大的实际意义。

中文关键词： 钙干预；生物标志物；代谢组学；动态概率主成分分析

英文摘要： Bone Mineral Density (BMD) is the main indicator assess the effect of calcium intervention. Given its poor accuracy and sensitivity and expensive cost, BMD cannot be used to evaluate treatment effect and identify susceptive group in the early stage of intervention study. In this study, dynamic probabilistic principal components analysis (DPPCA) will be fitted to screen significant biomarkers, which were measured in each time point. Then metabolic pathways will be explored. The main study contents are following: constructing DPPCA and simulation experiment, estimating model parameters, estimating time effect and testing significant biomarkers, data visualization, identifying biomarkers and biological explanation and so on. In this study, DPPCA will be constructed to analyze longitudinal metabolomics data and approach estimating model parameters will be provided. DPPCA can assess the effect of time within each treatment group and identify metabolites which change over time within each treatment group. This study will provide effective and reliable information to identify susceptive biomarkers of calcium supplementation trials. Thus susceptive group will be screened as early as possible. Therefore, it is important to modify strategy of calcium supplementation, population health and disease prevention.

英文关键词： calcium intervention;biomaker;metabonomics;dynamic probabilistic principal component analysis