BACKGROUND The microbial communities populating human and natural environments have been extensively characterized with shotgun metagenomics, which provides an in-depth representation of the microbial diversity within a sample. Microbes thriving in urban environments may be crucially important for human health, but have received less attention than those of other environments. Ongoing efforts started to target urban microbiomes at a large scale, but the most recent computational methods to profile these metagenomes have never been applied in this context. It is thus currently unclear whether such methods, that have proven successful at distinguishing even closely related strains in human microbiomes, are also effective in urban settings for tasks such as cultivation-free pathogen detection and microbial surveillance. Here, we aimed at a) testing the currently available metagenomic profiling tools on urban metagenomics; b) characterizing the organisms in urban environment at the resolution of single strain and c) discussing the biological insights that can be inferred from such methods. RESULTS We applied three complementary methods on the 1614 metagenomes of the CAMDA 2017 challenge. With MetaMLST we identified 121 known sequence-types from 15 species of clinical relevance. For instance, we identified several Acinetobacter strains that were close to the nosocomial opportunistic pathogen A. nosocomialis. With StrainPhlAn, a generalized version of the MetaMLST approach, we inferred the phylogenetic structure of Pseudomonas stutzeri strains and suggested that the strain-level heterogeneity in environmental samples is higher than in the human microbiome. Finally, we also probed the functional potential of the different strains with PanPhlAn. We further showed that SNV-based and pangenome-based profiling provide complementary information that can be combined to investigate the evolutionary trajectories of microbes and to identify specific genetic determinants of virulence and antibiotic resistances within closely related strains. CONCLUSION We show that strain-level methods developed primarily for the analysis of human microbiomes can be effective for city-associated microbiomes. In fact, (opportunistic) pathogens can be tracked and monitored across many hundreds of urban metagenomes. However, while more effort is needed to profile strains of currently uncharacterized species, this work poses the basis for high-resolution analyses of microbiomes sampled in city and mass transportation environments. REVIEWERS This article was reviewed by Alexandra Bettina Graf, Daniel Huson and Trevor Cickovski.

中文翻译：

---

使用宏基因组测序数据分析城市环境中的微生物菌株。

背景技术已经用shot弹枪宏基因组学广泛地描述了居住在人类和自然环境中的微生物群落，这提供了样品中微生物多样性的深入表示。在城市环境中蓬勃发展的微生物对于人类健康可能至关重要，但与其他环境相比，受到的关注较少。正在进行的努力开始大规模地针对城市微生物群落，但是用于描述这些元基因组的最新计算方法从未在这种情况下应用。因此，目前尚不清楚这样的方法是否已被证明成功地区分了人类微生物群系中甚至密切相关的菌株，在城市环境中是否也可以有效地用于无栽培病原体检测和微生物监测等任务。这里，我们的目标是：a）在城市宏基因组学上测试当前可用的宏基因组分析工具；b）以单一菌株的分辨率表征城市环境中的生物，并且c）讨论可以从这种方法中推断出的生物学见解。结果我们对CAMDA 2017挑战的1614个基因组应用了三种互补方法。通过MetaMLST，我们从15种具有临床意义的物种中鉴定出121种已知的序列类型。例如，我们确定了几株与医院机会病原体医院线虫接近的不动杆菌菌株。使用StrainPhlAn（MetaMLST方法的通用版本），我们推断了斯氏假单胞菌菌株的系统发育结构，并建议环境样品中的菌株水平异质性高于人类微生物组。最后，我们还使用PanPhlAn探索了不同菌株的功能潜力。我们进一步表明，基于SNV和基于基因组分析的概况分析提供了补充信息，可以将其组合起来以调查微生物的进化轨迹，并确定密切相关菌株内毒力和抗生素抗性的特定遗传决定因素。结论我们表明，主要用于分析人类微生物群的菌株水平方法对于城市相关微生物群可能是有效的。实际上，可以在数百个城市元基因组中跟踪和监视（机会性）病原体。但是，尽管需要付出更多的努力来描述当前尚未表征的物种的菌株，但这项工作为高分辨率分析城市和大众运输环境中采样的微生物群落奠定了基础。