III: Medium: Algorithms and Software Tools for Epigenetics Research

该项目将开发一个新的计算框架，以促进对人类疟疾寄生虫表观遗传基因调控的理解。 表观遗传学是研究基因表达或细胞表型的可遗传变化，这些变化是由基础DNA序列变化以外的机制引起的。计算框架的核心是解决一组硬计算问题的能力，这是研究计划的重点。计算的挑战需要研究新的组合优化问题，开发新的时间和空间效率的算法，并最终实现和部署用户友好的基于Web的软件工具。分析人类疟疾寄生虫表观基因组的能力将提高我们对其生物学的理解，并可能使分子生物学家能够确定新的抗疟策略。 拟议的计算框架也将使生命科学家作出新的表观遗传学的发现，并最终提高驱动基因表达在其他真核生物的复杂机制的理解。软件工具将被置于公共领域，这将使全世界的研究人员和公众受益，并可能导致新的国际和工业合作。这个项目将在一个高度跨学科的环境中支持两名研究生和一名博士后。大多数真核生物基因组都有第二层信息，这些信息嵌入在添加到DNA和特殊蛋白质的突出尾部的化学标记上，这些蛋白质将DNA包装成称为核小体的复合物。在过去的几十年里，分子生物学中最令人惊讶的发现之一就是这种被称为表观基因组的“隐蔽”层，它影响着各种细胞和代谢过程。表观遗传标记不仅控制着每种细胞中哪些基因是可接近的，而且还决定了可接近的基因何时被激活。分子生物学家还证实，表观基因组受到生物体与环境相互作用的影响，这些相互作用引起的表观遗传标记的变化在细胞分裂中遗传，尽管不是直接编码在DNA中。这个项目将研究一系列的计算挑战，将带来越来越多的表观基因组计划。具体而言，目标是开发用于以下的方法和软件工具：（1）分析核小体和甲基化图（使用修改的高斯混合模型和期望最大化）;（2）研究核小体定位、组蛋白尾部修饰和DNA甲基化模式的动力学（使用图论方法，例如，k-partite matching）;（3）稳定核小体和特异性组蛋白修饰的DNA基序分析（使用组合优化方法）;（4）使用核小体或甲基化景观发现新基因（使用机器学习分类器）;（5）核小体定位、组蛋白修饰、DNA甲基化模式和基因表达之间的统计学显著的全基因组相关性的鉴定（使用动态贝叶斯网络）。 这五项计算任务将需要研究新的组合优化和机器学习问题，开发新的时间和空间效率算法，并最终实现和部署用户友好的基于网络的软件工具。开发算法的“平台”是恶性疟原虫，每年造成3.5亿至5亿例疟疾病例的寄生虫，以及全球范围内100万到300万人的死亡。目前还没有针对疟疾的疫苗（一种疫苗目前正在临床试验中），这种寄生虫正在对几乎所有现有药物产生抗药性。开发的方法和工具将不是疟疾特异性的，并且将扩展到具有更大/更复杂基因组的各种其他真核生物。http://www.cs.ucr.edu/~stelo/iis13.htm

This project will develop a new computational framework to advance the understanding of epigenetic gene regulation in the human malaria parasite. Epigenetics is the study of heritable changes in gene expression or cellular phenotype caused by mechanisms other than changes in the underlying DNA sequence.At the core of the computational framework is the ability to solve a set of hard computational questions, which are the focus of the research plan. The computational challenges require the study of novel combinatorial optimization problems, the development of new time- and space-efficient algorithms, and ultimately the implementation and deployment of user-friendly web-based software tools. The ability to analyze the epigenome of the human malaria parasite will improve our comprehension of its biology and possibly enable molecular biologists to identify new antimalarial strategies. The proposed computational framework will also enable life scientists to make novel epigenetic discoveries and ultimately improve the understanding of the complex mechanisms that drive gene expression inother eukaryotic organisms. Software tools will be placed into the public domain, which will benefit researchers and the public worldwide, and potentially lead to new international and industrial collaborations. This project will support two graduate students and one post-doc in a highly interdisciplinary environment.Most eukaryotic genomes have a second layer of information which is embedded on chemical marks added to DNA and to the protruding tail of special proteins that package DNA into a complex called the nucleosome. One of the most astonishing discoveries in molecular biology of the past decades is that this "covert" layer, called the epigenome, affects a variety of cellular and metabolic processes. Epigenetic marks not only controls what genes are accessible in each type of cell, but also determine when the accessible genes may be activated. Molecular biologists have also confirmed that the epigenome is affected by the interactions of the organism with the environment and that changes to the epigenetic marks induced by these interactions are inherited across cell division, despite not being encoded directly in DNA. This project will study a set of computational challenges that will be brought about by the increasing number of epigenome projects. Specifically, the goal is to develop methods and software tools for (1) the analysis nucleosome and methylation maps(using a modified Gaussian mixture model and expectation maximization); (2) the study of dynamics of nucleosome positioning, histone tail modifications and DNA methylation patterns (using graph theoretical approaches, e.g., k-partite matching); (3) the analysis of DNA motifs for stable nucleosomes and specific histone modifications (using combinatorial optimization approaches); (4) the discovery of new genes using nucleosome or methylation landscapes (using machine learning classifiers); (5) the identification of statistically significant genome-wide correlations between nucleosome positioning, histone modifications, DNA methylation patterns and gene expression (using dynamic Bayesian networks). These five computational tasks will require the study of novel combinatorial optimization and machine learning problems, the development of new time- and space-efficient algorithms, and ultimately the implementation and deployment of user-friendly web-based software tools.The "platform" on which the algorithms will be developed is P. falciparum, the parasite responsible each year for 350-500 million cases of malaria, and between one and three million of human deaths world-wide. There is no vaccine against malaria (one is currently on clinical trials) and the parasite is developing resistances to almost all drugs currently available. The methods and tools developed will not be malaria-specific, and will scale to a variety of other eukaryota with much larger/complex genomes.Updates and additional information about this project will be made available at http://www.cs.ucr.edu/~stelo/iis13.htm