Multivariate Analysis of Candidate Blood Pressure Response Genes in Hypertensives

描述(由申请人提供)： 这个项目的主要目标是反向工程基因变异(在影响肾小管钠转运的候选基因中，并参与肾素-血管紧张素-醛固酮系统)和其他变量之间的相互作用，这些变量导致个体之间对噻嗪类利尿剂的血压反应的差异。我们将应用“途径”或“系统生物学”的数据分析方法来研究585名高血压患者对氢氯噻嗪的血压反应。此前，他们已经对16个候选基因的遗传变异进行了基因分型，并测量了一些中间表型和其他变量。已经进行了一系列单变量分析，并确定了少数具有统计学意义的血压反应预测因子(以及某些其他感兴趣的中间表型)。然而，这些分析的范围非常有限。我们打算继续进行多变量分析，以获得对数据集和潜在生物学途径的系统了解。我们打算将创新的多变量分析方法应用于数据，主要是贝叶斯网络(BN)建模和增强型分类器/聚类器。这种数据挖掘方法在应用于其他领域的相似数据集时已经取得了非常成功的效果。最后，我们将使用这一数据集在遗传流行病学的背景下确认BN建模在敏感性和稳健性方面的实用价值。我们还将研究应用于遗传数据的BN重建的各种技术方面。

DESCRIPTION (provided by applicant): The primary goal of this project is to reverse-engineer the interplay of gene variation (in candidate genes influencing renal tubular sodium transport and involved in renin-angiotensin-aldosterone system) and other variables contributing to the interindividual differences in blood pressure (BP) response to a thiazide diuretic. We will apply "pathway", or "systems biology", data analysis methods to the association study of BP response to a hydrochlorothiazide in 585 hypertensive individuals. Previously, they have been genotyped for genetic variation in 16 candidate genes, and measured for a number of intermediate phenotypes and other variables. A series of univariate analyses have been carried out, and a small number of statistically significant predictors of blood pressure response (and certain other intermediate phenotypes of interest) have been identified. However, the scope of these analyses was very limited. We intend to follow up with the multivariate analysis in order to gain the systemic understanding of the dataset and the underlying biological pathways. We intend to apply innovative multivariate analysis methods, predominantly Bayesian Network (BN) modeling and boosted classifiers / clusterers, to the data. Such data mining methods have been very successful when applied to the similar datasets in other domains. Finally, we will use this dataset to confirm, in the context of genetic epidemiology, the practical utility of BN modeling with respect to sensitivity and robustness. We will also investigate various technical aspects of BN reconstruction, as applied to the genetic data.