Dipartimento di Fisica "G. Galilei" e INFN sezione di Padova, University of Padova, Padova, Italy.
Dipartimento di Scienze Chirurgiche, Oncologiche e Gastroenterologiche (DiSCOG), University of Padova, Padova, Italy.
PLoS Comput Biol. 2024 Sep 27;20(9):e1012482. doi: 10.1371/journal.pcbi.1012482. eCollection 2024 Sep.
Recent advancements in next-generation sequencing have revolutionized our understanding of the human microbiome. Despite this progress, challenges persist in comprehending the microbiome's influence on disease, hindered by technical complexities in species classification, abundance estimation, and data compositionality. At the same time, the existence of macroecological laws describing the variation and diversity in microbial communities irrespective of their environment has been recently proposed using 16s data and explained by a simple phenomenological model of population dynamics. We here investigate the relationship between dysbiosis, i.e. in unhealthy individuals there are deviations from the "regular" composition of the gut microbial community, and the existence of macro-ecological emergent law in microbial communities. We first quantitatively reconstruct these patterns at the species level using shotgun data, and addressing the consequences of sampling effects and statistical errors on ecological patterns. We then ask if such patterns can discriminate between healthy and unhealthy cohorts. Concomitantly, we evaluate the efficacy of different statistical generative models, which incorporate sampling and population dynamics, to describe such patterns and distinguish which are expected by chance, versus those that are potentially informative about disease states or other biological drivers. A critical aspect of our analysis is understanding the relationship between model parameters, which have clear ecological interpretations, and the state of the gut microbiome, thereby enabling the generation of synthetic compositional data that distinctively represent healthy and unhealthy individuals. Our approach, grounded in theoretical ecology and statistical physics, allows for a robust comparison of these models with empirical data, enhancing our understanding of the strengths and limitations of simple microbial models of population dynamics.
近年来，下一代测序技术的进步彻底改变了我们对人类微生物组的理解。尽管取得了这些进展，但在理解微生物组对疾病的影响方面仍然存在挑战，这是由于物种分类、丰度估计和数据组合性方面的技术复杂性所导致的。与此同时，最近使用 16s 数据提出了描述微生物群落变异和多样性的宏观生态法则的存在，这些法则不受环境的影响，并通过一个简单的种群动态现象学模型来解释。我们在这里研究了肠道微生物群落的失调（即不健康个体的组成与“正常”组成存在偏差）与微生物群落中宏观生态突现法则之间的关系。我们首先使用 shotgun 数据在物种水平上定量重建这些模式，并解决采样效应和统计误差对生态模式的影响。然后，我们询问这些模式是否可以区分健康和不健康的队列。同时，我们评估了不同的统计生成模型的功效，这些模型包含了采样和种群动态，可以描述这些模式，并区分哪些是随机出现的，哪些可能与疾病状态或其他生物学驱动因素有关。我们分析的一个关键方面是理解模型参数与肠道微生物组状态之间的关系，从而能够生成独特代表健康和不健康个体的合成组成数据。我们的方法基于理论生态学和统计物理学，允许与经验数据进行稳健比较，从而增强我们对简单微生物种群动态模型的优势和局限性的理解。
