Quantitative Modeling of Sequence-to-Expression Relationship

 描述(由申请人提供)：了解基因是如何开启和关闭的，以及它们的精确表达水平是如何受到调控的，这对于描述基因变异与人类健康之间的联系至关重要。正在进行的社区范围的努力承诺对各种条件下的基因组状态的海量信息(数据)进行编目，包括特定的疾病状态。这类目录有望帮助确定特定条件基因表达的关键调控因素。然而，“读取”DNA序列并准确预测任何特定细胞的表达水平的终极梦想可能仍然遥不可及。我们建议开发先进的计算工具，帮助生物学家和基因组科学家实现从序列预测基因表达水平的最终目标。这项提议的第一个也是主要的目标是建立一个软件系统，帮助生物学家对基因表达与调控序列的关系进行建模。在这里，“模型”指的是用数量语言描述序列和表达之间的关系，具有非常高的准确性。拟议的软件系统将被称为‘GEM’(基因表达建模)，它将巩固我们在过去五年中在这一方向上的努力，并将新的生化方面融入到模型中。与这一领域的标准不同，拟议的软件将向生物学家展示与收集的数据一致的所有模型，而不仅仅是最令人满意的单一模型。换句话说，科学家将看到他们的数据在细胞内的基因调控相互作用方面的所有可能的解释。第二个目标是致力于以易于理解的格式向科学家展示模型，包括各种视觉表示。这里的目标是将上述模型典型的量化和抽象形式与生物学家对基因调控机制的更切实的概念联系起来。这项提议的第三个目的是帮助生物学家改进在目标1中创建的模型，要么通过假设到目前为止未知的基因调节器的存在，要么通过产生额外的数据。该软件系统将使用严格的统计方法和客观标准来帮助生物学家决定哪些实验应该是最有成效的，以促进他们对基因调控系统的理解。所有的具体目标都将在昆虫和哺乳动物的四个重要调控系统上进行评估。

 DESCRIPTION (provided by applicant): Understanding how genes are turned on and off, and how their precise levels of expression are regulated, is critical to describing the connection between genetic variations and human health. On-going community-wide efforts promise to catalog vast amounts of information (data) about genomic states in a variety of conditions, including specific disease states. Such catalogs are expected to help identify key regulators of condition-specific gene expression. However, the ultimate dream of `reading' the DNA sequence and accurately predicting expression levels in any given cell is likely to remain elusive. We propose to develop advanced computational tools that will help biologists and genome scientists realize this final goal of predicting gene expression levels from sequence. The first and main goal of this proposal is to build a software system that will help a biologist model how gene expression relates to regulatory sequences. Here, `model' refers to describing the relationship between sequence and expression in a quantitative language, with a very high level of accuracy. The proposed software system, to be called `GEM' (Gene Expression Modeling), will consolidate our efforts in this direction for the last five years, and also incorpoate novel biochemical aspects to the model. In a departure from the norm in this field, the proposed software will present to the biologist all models consistent with the collected data, and not just the single most agreeable model. In other words, the scientist will get to see all possible interpretations of their data in terms of gene regulatory interactions in the cell. The second aim is devoted to presenting the model to the scientist in easily interpretable formats, including a variety of visual representations. The goal here is to connect the typically quantitative and abstract form of the above-mentioned models to the more tangible notions the biologist has about gene regulation mechanisms. The third aim of this proposal is to help the biologist improve the models created in Aim 1, either by hypothesizing the existence of hitherto unknown regulators of the gene, or by generating additional data. The software system will use rigorous statistical methods and objective criteria to help the biologist decide which experiments should be most productive in advancing their understanding of the gene regulatory system. All specific aims will be evaluated on four important regulatory systems from insects and mammals.