Zhang Yilong, Han Sung Won, Cox Laura M, Li Huilin
Merck Research Laboratories, Rahway, New Jersey, United States of America.
Fusion Data Analytics Lab, School of Industrial Management Engineering, Korea University, Seoul, South Korea.
Genet Epidemiol. 2017 Dec;41(8):769-778. doi: 10.1002/gepi.22065. Epub 2017 Sep 5.
Human microbiome is the collection of microbes living in and on the various parts of our body. The microbes living on our body in nature do not live alone. They act as integrated microbial community with massive competing and cooperating and contribute to our human health in a very important way. Most current analyses focus on examining microbial differences at a single time point, which do not adequately capture the dynamic nature of the microbiome data. With the advent of high-throughput sequencing and analytical tools, we are able to probe the interdependent relationship among microbial species through longitudinal study. Here, we propose a multivariate distance-based test to evaluate the association between key phenotypic variables and microbial interdependence utilizing the repeatedly measured microbiome data. Extensive simulations were performed to evaluate the validity and efficiency of the proposed method. We also demonstrate the utility of the proposed test using a well-designed longitudinal murine experiment and a longitudinal human study. The proposed methodology has been implemented in the freely distributed open-source R package and Python code.
人类微生物组是生活在我们身体各个部位内外的微生物集合。自然界中生活在我们身体上的微生物并非独自存在。它们作为一个整合的微生物群落，有着大量的竞争与合作，并以非常重要的方式对我们人类的健康做出贡献。目前大多数分析集中在检查单个时间点的微生物差异，这无法充分捕捉微生物组数据的动态本质。随着高通量测序和分析工具的出现，我们能够通过纵向研究探究微生物物种之间的相互依存关系。在此，我们提出一种基于多元距离的检验方法，利用重复测量的微生物组数据来评估关键表型变量与微生物相互依存之间的关联。进行了广泛的模拟以评估所提出方法的有效性和效率。我们还通过精心设计的纵向小鼠实验和纵向人类研究证明了所提出检验方法的实用性。所提出的方法已在免费分发的开源R包和Python代码中实现。
