【作者】 舒文杰；

【导师】 郑志强； 王升启；

【作者基本信息】 国防科学技术大学 ， 控制科学与工程， 2007， 博士

【摘要】 核糖核酸(RNA)是一类重要的生物大分子,在生命活动中发挥着重要的作用。对RNA分子的研究不仅具有重要的理论和应用价值,而且还将导致对生命本质问题的深入认识。在此背景下,Filipowicz于2000年提出了RNA组学的概念。RNA组学的兴起对RNA高级结构的分析方法提出了新的要求,其中特别需要发展RNA二级结构的量化方法,即RNA二级结构的拓扑特征化方法。本文建立了RNA二级结构的拓扑特征化方法,并应用于RNA构象空间的可视化、RNA有害突变预测、RNA结构稳健性的定量评估、多稳态RNA分子设计,以及RNA进化动力学模拟,其主要内容和创新之处包括:(1)RNA二级结构拓扑特征化方法的研究RNA二级结构拓扑特征化方法的目的是构建能够定量表征RNA二级结构特征的拓扑指数,它是最近几年RNA结构分析领域新兴的研究方向。为了给RNA二级结构提出一个完备且精细的模型,本文首先发展了RNA树的模型,提出了RNA二级结构的茎环元件连接图的表示方法。基于这种新的表示,构建了三种表征能力强、简并度低的拓扑指数,并分析了它们的适用性。随后,在RNA二级结构山形图表示的基础上,定义了RNA二级结构的分形维指数。并对RNA二级结构的弱自相似性(WSA)进行了深入分析,建立了基于WSA函数表示的RNA分子的功能域识别的方法。(2)RNA二级结构拓扑特征化方法的基本应用研究前面构造的RNA二级结构拓扑指数为RNA构象空间的定量描述奠定了基础,但在应用中,可能会造成“维数灾难”的问题。为此,本文将流形学习用于RNA构象空间的可视化,建立了以拓扑指数为特征、基于流形学习降维的RNA构象空间的可视化方法,将该方法应用于交互式的miRNA识别与聚类,以及非编码RNA聚类中,取得了良好的效果。同时,本文提出了两种RNA有害突变预测的方法,一种是基于结构差异的方法;另外一种是基于结构差异与多序列序列比对的方法。并实现了RNA有害突变预测与分析的Web计算。最后,将RNA有害突变预测方法应用于流感病毒减活疫苗的设计,取得了良好的效果。(3)RNA结构稳健性分析与多稳态RNA分子设计研究稳健性的定量评估一直是进化生物学家长期关注的一个问题,而RNA二级结构是研究稳健性的一个很好的平台。本文将中性值的概念引入到RNA结构元件稳健性的定量评估中,建立了RNA结构元件遗传稳定性的定量评估方法,并以miRNA为研究对象,分析了miRNA的结构元件遗传稳健性。同时,本文扩展了中性值的定义,建立了RNA二级结构遗传稳定性的定量评估方法,同样以miRNA为研究对象,分析了miRNA二级结构的遗传稳健性。并讨论了miRNA遗传稳健性的进化及其起源,提出了miRNA的遗传稳健性与环境稳健性协同进化的假设。为了方便RNA结构稳健性的研究,本文实现了其Web计算。本文在现有双稳态分子设计理论及其算法框架下,将多稳态RNA分子设计问题转化为与预定结构相容的RNA序列集上的组合优化问题,采用基于图论的数学模型,该问题可以简化为依赖图上的点着色问题。本文解决了多稳态RNA分子的可设计性问题,并提出了多稳态RNA分子设计的算法。在优化设计的代价函数中,除了热力学性质之外,还融入了RNA稳健性的信息。双稳态和多稳态RNA分子设计算例的结果表明所设计出的RNA序列很好地满足了设计要求。(4)RNA进化动力学模拟及其理论建模研究RNA分子进化动力学模拟及其理论建模一直是理论生物学家长期关注的一个问题。本文采用蒙特卡罗方法模拟RNA分子进化过程,建立了RNA分子进化的结构和自由能景观图。以miRNA为研究对象,分析了结构和自由能景观图的自相关性。利用高斯混合模型及其相关算法,本文建立了RNA景观图的理论模型。为了方便RNA进化动力学的研究,本文实现了其Web计算。更多还原

【Abstract】 As one of the most important macromolecules,Ribonucleic Acid(RNA) plays various indispensable roles in life.Researches on RNA molecules will result in deep recognitions of the core problems in the life science,as well as important applications. RNomics,which aims to reveal sequences and structure of all types of RNAs,has been proposed by Filipowicz in 2000.The emergence of RNomics present new requests for new methods for computational analysis of RNA high structure,topologization of RNA secondary structure,i.e.quantitative methods of RNA secondary structure are especially required.Some new methods for RNA secondary structure topologization are presented in this dissertation and applied to many researches,including visualization of RNA configuration space,RNA deleterious mutation prediction,quantitative evaluation of RNA structural robustness,multistable RNA molecules design,and RNA evolutionary dynamics.The main contents and contributions of the dissertation are summarized as follows:(1) RNA secondary structure topologizationTopologization of RNA secondary structure aims at constructing topological indices which can describe RNA secondary structure quantitatively.RNA secondary structure topologization is a newly emerging direction in the field of RNA structural analysis.To provide a complete and fine scheme for RNA secondary structure,a combination of three vertex-weighted element-contact graphs(ECGs) is first proposed. Both the stem and loop topologies are completely encoded in ECG groups.Three typical topological index families defined on ECGs are investigated.The numerical features of these indices for possible RNA topologies are explored.Fractal dimensions based on mountain representation of RNA secondary structure are also presented.Using the weakly self-affine(WSA) model of RNA secondary structure,a functional domain recognition method is presented in this dissertation.(2) Basic applications of RNA secondary structure topologizationThe RNA topological indices establish the foundation for the quantitative description of RNA configuration space.However,the use of these indices may result in "Dimension Disaster" in applications.A method of visualization of RNA configuration space based on manifold learning is presented,where RNA secondary structures are quantitatively described by these bopological indices.The method is successfully applied in interactive recognition and clustering of miRNAs,and clustering of ncRNAs.Two methods of RNA deleterious mutation prediction are proposed.One is based on the structural difference,and the other is based on the structural difference and multiple sequence alignment.Also,a web server of RNA deleterious mutation analysis is implemented.These methods are successfully applied in the design of LAIV(Live Attenuated Influenza Vaccine).(3) RNA structural robustness and muitistable RNA molecules designEvolutionary biologists have a long-standing interest in quantitative measure of robustness.Many merits of RNA secondary structure make them an ideal system to study robustness.A method of quantitatively measuring robustness of structural elements based on neutrality is first developed,and the robustness of structural elements in miRNAs is analyzed using this method.This dissertation presents a method of quantitatively measuring robustness of RNA secondary structure,and investigates the structural robustness of miRNAs.Furthermore,this dissertation discusses the evolution of genetic robustness in miRNAs,and proposes the hypothesis of congruent evolution of genetic and environmental robustness.Finally,a web server of RNA structural robustness evaluation is implemented for the study of robustness.Based on the theory and algorithm of the design of bistable RNA molecules,the design of multistable RNA molecules can be described as the combinatorial optimization problem on the set of compatible sequences.Using the graph theory,this problem can be reduced into a vertex coloring problem of dependency graph.An algorithm for multistable RNA molecular design is presented,where the cost function involves the information of thermodynamic stability and genetic robustness of RNA molecules.The designed RNA sequences satisfy the requests well.(4) Simulation of RNA evolutionary dynamics and its theory modelTheoretical biologists have paid close attention to the study of simulation and theoretical model of RNA evolutionary dynamics.Using Monte Carlo method,this dissertation simulates the process of RNA evolution,and establishes the corresponding structural and free energy landscape.The landscapes of miRNAs have been analyzed in detail.Furthermore,a theoretical model of RNA landscape is developed based on Gaussian mixture model.Finally,a simulation platform is implemented for the study of RNA evolutionary dynamics.更多还原