微生物组大数据整合与数据挖掘方法研究

Microbes are ubiquitous on earth, and they usually live in the form of communities. Researches in microbiome, which include all genotypic and phenotypic information for the microbial community, would have profound important for understanding of functions, dynamics and interactions of microbial community, as well as their responses and feed-backs for the environments. Microbiome research also has applications in a wide-range of areas such as bio-medicine and environment monitoring. However, microbiome data possesses several properties, including multi-source, heterogeneity and big-data. Yet current methods are not fully capable of analyzing and mining these microbiome big-data, making the understanding of the structural and functional profiles for microbial communities difficult. Therefore, integrated and thorough analysis approach should be taken to fully understand the structure, function, interaction and dynamics of microbial communities...In this work, we plan to conduct research on "Method development for integration and data-mining for microbiome data". We will design taxonomical and functional data models to integrate multi-source and multi-type data from microbiome samples, including metagenomic data, metabolomic data and community meta-data. We will also design high-performance data-mining methods, including community structure and function profiling, species network analysis, and community-environment correlation analysis, for in-depth examination and understanding of microbial communities. This integrated and thorough analysis approach focusing on big-data mining and explanation for microbial community would deepen our understanding of the microbial communities’ structures, functions, as well as their adaptations and feedbacks for the environments, which would also have potential for a wide-range of application areas.

微生物在自然界中普遍存在并主要以“微生物群落”的形式共存。对包括群落基因型和表型等大量信息在内的微生物组的深入分析，允许我们更好的认识群落的结构和功能，群落对于生态环境的响应及反馈，以及微生物之间的相互影响和调控机制，发掘群落重要功能和在健康与环境等领域的潜在应用途径。然而微生物组数据具有多来源、异质性、数据量大等特点，导致现有分析方法无法全面、深入、高效的解析微生物组大数据。.本项目聚焦于微生物组大数据分析，计划整合包括元基因组数据、群落代谢数据和群落环境参数在内的海量微生物组数据，构建基于物种、功能、代谢产物、环境因素等特征的群落结构和功能数据模型。同时计划设计大数据挖掘算法，研究群落结构和功能特征、群落内部物种相关性网络、群落和环境关联性等，实现大数据驱动的微生物群落数据挖掘和理解。项目的开展有助于加深我们对微生物群落结构、功能、环境响应与反馈等重要问题的理解，促进相关应用研究开展。

微生物在自然界中普遍存在并主要以“微生物群落”的形式共存，其生存和演化受到生态环境的影响并对环境有重要的反馈作用。对微生物群落相关信息（微生物组）的深入分析，允许我们从新的层面认识和理解群落对于广义生态环境的响应及反馈，以及微生物物种之间的相互影响和调控机制，发掘群落微生物重要功能和在健康与环境等领域的潜在应用途径。然而微生物群落数据具有多来源、异质性、数据量大等特点，现有分析方法无法全面、深入、高效的解析微生物组大数据。..本项目“微生物组大数据整合与数据挖掘方法研究”的开展，取得了若干微生物数据挖掘方面的成果：.首先，本项目计划中涉及到的微生物组大数据整合与数据挖掘方法的开发，已完成微生物组样本搜索算法Meta-Prism [Briefings in Bioinformatics, 2020]，微生物组样本中亚种识别算法[Bioinformatics, 2019]等。同时开发了基于微生物组功能基因挖掘的蛋白结构和功能预测方法[Genome Biology, 2019]。.其次，本项目计划中涉及到的数据分析平台的开发，已开发微生物群落元基因组大数据研究的计算平台和开放式网站SEK，目前正在逐步整合进入国家组学数据百科全书NODE系统（https://www.biosino.org/node/）。.最后，本项目计划中涉及到的微生物组大数据整合与数据挖掘方法的应用，已完成肠道菌群时空可塑性研究[Gut, 2019]，运动员肠道微生物群落模式的挖掘[Gut Microbes, 2020]等多项应用研究，发掘了一系列人体肠道菌群变化模式。..项目的开展推动了对群落更为透彻的分析，加深了我们对微生物群落结构、功能、环境反馈等重要问题的理解，进而促进了相关应用研究开展。