A Unified Computational Model of Gene Regulation

理解存储在DNA中的指令如何以不同的方式被解释以产生不同的细胞活动是我们这个时代最基本的、未解决的问题之一。这种“生命软件”可以解释不同个体、后代和物种之间的关系。它将告诉我们基因突变如何导致表型或疾病的变化。它将提供生物分子疾病治疗效果的预测。它将帮助我们在细胞层面上理解我们是谁。最近，一个由全球数千名基因组研究人员组成的联盟发现了大量的DNA元素，这些元素似乎对控制细胞内的活动很重要（ENCODE, 2012）。我们现在知道多达80%的人类基因组包含这些重要的DNA元素，但知道一些重要的东西并不等于了解它是如何工作的。这是我们研究项目的主题。在这个项目中，我们的目标是建立一个“调控模型”，解释DNA元素如何结合在一起来控制细胞用来控制大多数活动的遗传信息（mrna）。包括我在内的几个研究小组在过去的10年里一直在研究这个问题，但进展一直很困难。虽然我们现在有许多描述基因组及其遗传信息的大规模数据集，但我们缺乏能够有效地将DNA特征与细胞条件描述结合起来预测遗传信息的良好计算模型。我们将研究一种新的计算模型，它统一并概括了以前的工作，这有助于我们理解可测试的生物机制，可以用来预测新基因和细胞状态的遗传信息水平，这是基于定义良好的，可测试的原则。虽然这是一个高风险的项目，但潜在的影响是巨大的，我们的初步结果表明，我们有充分的理由相信我们会成功。

Understanding how instructions stored in DNA are interpreted differently to generate diverse cellular activities is one of the most fundamental, unsolved problems of our time. This "software of life" would explain relationships between different individuals, offspring and species. It would tell us how genetic mutations lead to changes in phenotype or disease. It would offer predictions of the effects of biomolecular disease treatments. It would help us understand, at the cellular level, who we are.Recently, a consortium of thousands of genome researchers from across the globe identified a huge number of DNA elements that appear to be important for controlling activities within cells (ENCODE, 2012). We now now that up to 80% of the human genome contains these important DNA elements, but knowing that something is important is not the same as understanding how it works. This is the topic of our research project. In this project, our goal is to build a "regulatory model" that explains how DNA elements combine together to control the genetic messages (mRNAs) that cells use to control most activites. Several research groups, including mine, have been studying this problem for the past 10 years, but progress has been difficult. While we now have many massive datasets that profile genomes and their genetic messages, we lack good computational models that can effectively combine DNA features with descriptions of cellular conditions to predict genetic messages. We will investigate a novel computational model that unifies and generalize previous work, that helps us understand testable biological mechanisms, that can be used to predict levels of genetic messages for novel genes and cell states, and that is based on well-defined, testable principles. While this is a high risk project, the potential impact is tremendous and our preliminary results indicate that there is good reason to believe we will succeed.