Statistical Methods for High-Throughput Gene Expression Profiling

描述（由申请人提供）：在全基因组水平上进行的研究现在在现代生物学和医学中发挥着核心作用。这些研究仍然迫切需要可应用的新统计方法，特别是随着研究设计变得更加雄心勃勃、样本量增加以及新技术出现。拟议研究的总体目标是开发有助于理解以表征全基因组基因表达为中心的高通量分子分析数据的统计方法和软件。我们建议开发统计模型、方法和软件，使人们能够在研究设计和潜在变量方面严格表征基因表达的变化。我们提出的研究特别关注最现代形式的基因表达谱、RNA-Seq，以及目前正在研究的最雄心勃勃且在生物学上富有成果的问题。我们将开发严格、灵活和稳健的高通量数据变异模型，其中包括：（i）系统变异的潜在来源和特征之间更普遍的依赖性来源，（ii）对下一代RNA-Seq数据变异来源的原则性剖析和新兴RNA-Seq数据的新方法，（iii）对基因组特征和潜在变量之间的关联进行严格的多重假设检验，（iv）同时推断复杂但普遍寻求的统计假设，以及（v）传播到更广泛的领域通过用户友好且独立于平台的软件包建立研究社区。

DESCRIPTION (provided by applicant): Studies carried out at the genome-wide level now play a central role in modern biology and medicine. There continues to be a substantial need for new statistical methods that can be applied in these studies, particularly as study designs become more ambitious, sample sizes increase, and new technologies emerge. The overall goal of the proposed research is to develop statistical methods and software useful in understanding high- throughput molecular profiling data centered around characterizing genome-wide gene expression. We propose to develop statistical models, methods, and software that allow one to rigorously characterize variation of gene expression in terms of both study design and latent variables. Our proposed research is particularly focused on the most modern form of gene expression profiling, RNA-Seq, as well as the most ambitious and biologically fruitful problems currently being studied. We will develop rigorous, flexible, and robust models of variation in high-throughput data that encompass: (i) latent sources of systematic variation and more general sources of dependence among features, (ii) a principled dissection of sources of variation in next-generation RNA-Seq data and new methods for emerging RNA-Seq data, (iii) rigorous multiple hypothesis testing of associations between genomics features and latent variables, (iv) simultaneous inference of complex, yet commonly sought after statistical hypotheses, and (v) dissemination to the greater research community through user-friendly and platform independent software packages.