Applied Statistics to a Secondary Analysis of Public Repositories for Microarray

描述（由申请人提供）：全基因组基因表达谱的使用越来越多，产生了大量有价值的数据，这些数据提供了具有成本效益的第二次机会，以调查原始预期目的中未包括的其他研究问题。我们的目标是在坚实的统计基础上开发系统的方法，对现有的微阵列表达数据库进行二次分析。我们将强调生物背景和统计建模之间的一致性。这种一致性对于提高所开发的分析工具的生物学效率至关重要。视网膜是中枢神经系统的一个相对简单且特征明确的区域。目前，已经发现了200多个导致视网膜疾病的基因。我们将应用开发的方法，视网膜微阵列表达数据库，以确定新的基因和基因-基因关系（途径），管理正常和病理过程的视网膜。我们还将探索通过公共数据库搜索和比较进行眼病预测的可能性。我们提出以下具体研究：1）开发新的分析/统计方法来检测参与生物途径的基因。我们计划设计一个统计策略，将部分相关性作为该应用程序的核心组成部分; 2）迈出将微阵列库转化为疾病诊断数据库的第一步。我们计划开发一种贝叶斯概率方法来推断查询微阵列数据集的疾病状况，基于其与数据库中那些充分表征的数据的相似性; 3）实验验证硅预测的子集。我们将使用标准方法如实时PCR和Western印迹验证首次研究中新鉴定的基因的表达; 4）扩展我们现有的软件包Gene Expression Analyzer（GEA）（http：//cell.rutgers.edu/gea/）以包括新开发的方法。源代码将被公开。该项目的成果将大大促进大量公共数据集的重用，以回答更多的研究问题，减少生成新数据的必要性，并提高我们对各种扰动下细胞功能和网络的理解。

DESCRIPTION (provided by applicant): The increasing use of genome-wide gene expression profiling has generated great valuable data that offer cost-effective secondary opportunities to investigate additional research questions that were not included in the original intended purpose. Our goal is to develop systematic approaches on a firm statistical footing to conduct a secondary analysis of the existing microarray expression databases. We will lay emphasis on the consistency between the biological background and the statistical modeling in the developments. Such consistency is critical for enhancing the biological efficiency of the developed analysis tools. The retina is a relatively simple and well-characterized area of the central nervous system. Currently, over 200 genes were identified that cause retinal diseases. We will apply the developed methods to retinal microarray expression databases to identify novel genes and gene-gene relationships (pathways) that govern the normal and pathological processes of the retina. We will also explore the possibility of making eye disease predictions through a public database search and comparison. We propose the following specific studies: 1) to develop novel analytical/statistical methods for detecting the genes involved in a biological pathway. We plan to design a statistical strategy that incorporates partial correlation as a core component in this application; 2) to take the first step of turning microarray repositories into a disease diagnosis database. We plan to develop a Bayesian probabilistic method to infer the disease condition of a query microarray data set based on its similarity to those well-characterized data in database; 3) to experimentally validate a subset of in silicon predictions. We will verify the expression of newly identified genes from the first study using standard methods such as real-time PCR and Western blot; 4) to expand our existing software package Gene Expression Analyzer (GEA) (http://cell.rutgers.edu/gea/) to include the newly developed methods. The source code will be made public. The outcome of the project will significantly facilitate the reuse of the vast amount of public datasets to answer additional research questions, reduce the necessity to generate new data, and improve our understanding of cellular functions and networks under a variety of perturbations.