Compositional descriptors for large scale comparative metagenome analysis

宏基因组学和元转录组学为研究自然条件下微生物群落的系统发育分布、功能潜力和代谢活性提供了重要工具。目前，测序技术的快速发展极大地改善了宏基因组研究的基础。然而，不断增长的数量和序列材料的异质性，意味着生物信息学的一个巨大的挑战。该项目的目标是实现一个生物信息学框架，用于比较宏基因组和元转录组分析，该分析基于来自样品蛋白质结构域分布的统计描述符。这些描述符将被设计为在不同的测序平台之间提供高度的不变性，并提供关于微生物群落的系统发育和功能组成的高度信息化视图。将利用概率混合建模方法实现分类和功能描述符的跨平台可比性。通过包括由申请人开发的用于鉴定蛋白质结构域的超快速方法，该方法将容易地适用于任何规模的宏基因组项目，并且结果甚至将由不具有广泛计算设施的小型实验室和机构再现。为了评价这一方法，将与海洋微生物学领域的合作伙伴密切合作，在真实的世界条件下测试这一方法，为改进方法提供宝贵的反馈。

Metagenomics together with metatranscriptomics provides an essential tool for the investigation of the phylogenetic distribution, the functional potential and the metabolic activity of microbial communities under natural conditions. Currently, rapid progress in sequencing technologies dramatically improves the basis for metagenomic studies. The growing amount and the heterogeneity of sequence material, however, implies a big challenge for bioinformatics. The goal of this project is to realize a bioinformatics framework for comparative metagenome and metatranscriptome analysis which is based on statistical descriptors derived from the protein domain distribution of a sample. These descriptors will be designed to provide a high degree of invariance across different sequencing platforms and a highly informative view on the phylogenetic and functional composition of microbial communities. A probabilistic mixture modelling approach will be utilized to achieve cross-platform comparability of taxonomic and functional descriptors. With the inclusion of an ultra-fast method for identification of protein domains, developed by the applicant, the approach will readily be applicable to metagenomic projects of any size, and the results will even be reproducible by small labs and institutions which do not possess extensive computational facilities. For evaluation of the approach strong collaboration partners from the field of marine microbiology will be involved to test the methodology under real world conditions providing valuable feedback for refinement of the methods.