Breaking the Histone Code: Predicting Genome Organization with Chromatin States

人类基因组由46个DNA分子（染色体）组成，在完全延伸时可以延伸近两米，但必须适合直径只有10微米的细胞核。一个尚未解决的主要问题是基因组是如何在这个微小的三维空间中包装和组织的。该项目旨在通过开发新的理论和计算方法来描述人类基因组的三维组织来解决这个问题。一个类似谷歌图谱的基因组资源将被设计出来，这样人们就可以很容易地在多个长度尺度的层次组织中导航。这将被证明是研究与DNA基因组信息转化为构成细胞“主力”的RNA和蛋白质分子相关的遗传机制的宝贵援助。该项目还将为高中生、本科生、研究生和博士后提供衔接化学、物理和生物的优秀跨学科培训。本项目旨在建立一个结合统计力学、计算建模和生物信息学分析的综合框架的预测基因组模型。将开发新的理论方法来推导模型的两个关键组成部分，包括：(1)唯一代表来自给定细胞类型的染色体的输入序列；(2)描述势能面的力场，其全局最小值决定了最稳定的基因组结构。为了捕捉不同细胞类型的基因组构象的变化，表观遗传信息——包括反映“染色质状态”的组蛋白修饰——将被叠加在DNA序列之上，并用作输入。物理属性将被用来最大化信息熵。远程接触电位将使用严格的统计优化算法从现有的公开可用的全基因组染色体构象捕获数据中获得。这种预测性基因组模型不仅可以在广泛的细胞类型中实现基因组组织的高分辨率表征，而且还可以帮助揭示基因组折叠的潜在物理原理和驱动力。此外,该模型将为基础生物数据集（来自人类ENCODE项目）增加价值，并使基因组组织与功能输出相关的可测试假设得以形成。该奖项由生物科学理事会分子和细胞生物科学部的遗传机制项目和数学系的计算和数据支持科学与工程项目共同资助数学和物理科学理事会的科学。

The human genome, composed of 46 DNA molecules (chromosomes), stretches for nearly two meters when fully extended, yet must fit into a cellular nucleus that is only 10 micrometers in diameter. A major unsolved question is how the genome is packaged and organized in this tiny three-dimensional space. This project seeks to address this question by developing novel theoretical and computational approaches to characterize the three-dimensional organization of the human genome. A Google Map-like resource for the genome will be designed such that one can easily navigate through its hierarchical organization across multiple length scales. This will prove an invaluable aid for studying genetic mechanisms related to the translation of information in the DNA genome to the RNA and protein molecules that constitute the "workhorses" of the cell. The project will also provide excellent interdisciplinary training that bridges chemistry, physics and biology for high school students, undergraduates, graduate students and postdoctoral fellows.This project aims to build a predictive genome model with an integrative framework that combines statistical mechanics and computational modeling with bioinformatics analysis. Novel theoretical approaches will be developed to derive the two key components of the model, including (1) an input sequence that uniquely represents a chromosome from a given cell type; and (2) a force field that describes the potential energy surface whose global minimum determines the most stable genome structure. To capture the variation of genome conformation across cell types, epigenetic information--including histone modifications, which are reflective of "chromatin states"--will be superimposed on top of the DNA sequence and used as input. Physical attributes will be used to maximize the information entropy. Long-range contact potentials will be derived from existing, publicly available genome-wide chromosome conformation capture data using a rigorous statistical optimization algorithm. This predictive genome model will not only enable a high-resolution characterization of the genome organization across a wide range of cell types, but can also help uncover the underlying physical principles and driving forces for genome folding. Moreover, the model will add value to the foundational biological datasets (from the human ENCODE project) and enable formulation of testable hypotheses relating genome organization to functional outputThis award is co-funded by the Genetic Mechanisms Program in the Division of Molecular and Cellular Biosciences in the Biological Sciences Directorate and by the Program for Computational and Data-Enabled Science and Engineering in Mathematical and Statistical Sciences in the Division of Mathematical Sciences in the Mathematical and Physical Sciences Directorate.