Statistical assessment of complex and large networks derived from (meta)genomic data

在过去的十年中，测序技术的进步已经从包括原核生物和真核生物在内的无数生物中产生了大量的序列，并且随着成本的不断降低和序列读取的数量和长度的增加而实现了这一点。此外，元基因组学目前直接从环境或组织样本中提供了前所未有的丰富的DNA序列，可以用来详细描述生物系统的巨大复杂性、多样性和进化动力学。生物信息学资源已经成为组织、集成和比较这些海量数据的关键瓶颈。因此，生物学家只专注于对部分数据的分析，留下了大量未被探索的数据。这项建议的目标是开发一系列统计和图论方法，以使用基于网络的方法有效地分析大分子数据集。网络通常指的是一组离散实体及其在多维空间中的相互作用的基于数学图的表示。网络是分析生物系统及其组成部分之间复杂关系的有用概念工具；它们可以应用于基因组、元基因组和元翻译数据，其中单个实体(节点)可以由基因、基因类型、个人、物种、地理位置或这些组织级别的任何组合来表示。连接这些节点的链接(边)可以代表系统发育距离、遗传相似性或地理距离。如何从如此庞大的网络中提取有用的信息，并评估其跨越时间和空间的结构，是研究人员迫切感兴趣的问题。然而，我们缺乏统计框架来识别和比较不同微生物栖息地之间的这些网络。该方案旨在开发高效和强大的统计方法来解决大型网络数据集中的三类问题：(1)如何度量网络中组成节点的多样性；(2)如何比较复杂网络的拓扑模式；(3)如何使用网络分析来评估演化过程。也就是说，我打算设计新的多样性指数来表征巨型元基因组网络的复杂网状连接。然后，这些指数将被用于检测基因组网络中遗传分化的进化对象，并估计准确测量网络遗传多样性所需的采样工作量。我还将开发方法来有效地比较海量元基因组网络的拓扑。然后，这些方法将被应用于在不同时间收集的环境样本中检测一致的模式，并比较元基因组和元翻译网络。最后，我将提出统计零模型，用于测试代表不同生物系统的复杂网络中是否存在共同的进化过程。然后，这些统计模型将被用于检测基因组网络中的横向基因转移，并评估元基因组网络的可塑性。随着对各种基因组、元基因组和相互作用的数据集的应用，所提出的统计框架将对对网络动力学和时间和空间复杂性的研究感兴趣的研究人员具有重要的价值，特别是在人类微生物组中。这种新颖和稳健的基于网络的分子多样性及其时空变化的估计对于同时分析从单个病毒或细菌细胞到微生物群落的多个生物组织水平上的系统动力学和稳定性至关重要。

During the last decade, advances in sequencing technologies have generated an enormous volume of sequences from a myriad of organisms including both prokaryotes and eukaryotes, and have done so with continuous cost reduction and increased numbers and lengths of sequence reads. In addition, metagenomics is currently providing an unprecedented richness of DNA sequences directly from environmental or tissue samples that can be used to describe in detail the enormous complexity, diversity, and evolutionary dynamics of biological systems. Bioinformatics resources have become a critical bottleneck for the organization, integration, and comparison of these huge amounts of data. As a consequence, biologists focus on the analysis of only part of the data, leaving a large amount of data unexplored. The objective of this proposal is to develop a series of statistical and graph-theoretical methods to efficiently analyze large molecular datasets using network-based approaches. Networks generally refer to mathematical graph-based representations of a set of discrete entities and their interactions in multidimensional space. Networks are useful conceptual tools for analyzing biological systems and the complex relationships among their constituents; they can be applied to genomic, metagenomic, and metatranscriptomic data in which individual entities (nodes) can be represented by genes, genotypes, individuals, species, geographic locations, or any combination of such levels of organization. Links (edges) connecting such nodes can represent phylogenetic distances, genetic similarities, or geographic distances. How to extract useful information from such massive networks and assess their structure through time and across space is a question of acute interest for researchers. Yet we lack the statistical framework to identify and compare these networks between different microbial habitats. The proposal aims at developing efficient and powerful statistical approaches to address three types of problems in large network datasets: (1) how to measure the diversity of the constituent nodes in a network; (2) how to compare topological patterns of complex networks; and (3) how to assess evolutionary processes using network analysis. Namely, I intend to design novel diversity indices to characterize the complex reticulate connections of giant metagenomic networks. These indices will then be applied to detect genetically divergent evolving objects in genome networks, and to estimate the sampling effort required to accurately measure the genetic diversity of a network. I will also develop methods to efficiently compare topologies of massive metagenome networks. These methods will then be applied to detect congruent patterns in environmental samples collected at different times, and to compare metagenomic and metatranscriptomic networks. Finally, I will propose statistical null models for testing the presence of common evolutionary processes in complex networks that represent different biological systems. These statistical models will then be applied to detect lateral gene transfer in genome networks and to assess the plasticity of metagenome networks. With applications to various genomic, metagenomic and interactomic datasets, the proposed statistical framework will be of great value to researchers interested in the study of network dynamics and complexity in time and space, and in particular within the human microbiome. Such novel and robust network-based estimates of molecular diversity and its spatio-temporal variation will be crucial to simultaneously analyze system dynamics and stability at multiple levels of biological organization, from individual viruses or bacterial cells to microbial communities.