AUTHOR=Li Zhengbang , Yu Xiaochen , Guo Hongping , Lee TingFang , Hu Jiyuan 

TITLE=A maximum-type microbial differential abundance test with application to high-dimensional microbiome data analyses

JOURNAL=Frontiers in Cellular and Infection Microbiology

VOLUME=Volume 12 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/cellular-and-infection-microbiology/articles/10.3389/fcimb.2022.988717

DOI=10.3389/fcimb.2022.988717

ISSN=2235-2988

ABSTRACT=Background: High-throughput metagenomic sequencing technologies have shown prominent advantages over traditional pathogen detection methods, bringing great potential in clinical pathogen diagnosis and treatment of infection diseases.  Yet, how to accurately detect the difference of microbiome profiles between treatment or disease conditions remains computationally challenging. 

Results: In this study, we propose a novel test for identifying the difference between two high-dimensional microbiome abundance data matrices based on the centred log-ratio transformation of the microbiome compositions. The test p-value can be calculated directly with a closed-form solution from the derived asymptotic null distribution. We also investigate the asymptotic statistical  power against sparse alternatives which are typically encountered in microbiome studies. The proposed Maximum-type test is  Equal-Covariance-Assumption-Free (MECAF), making it widely applicable to studies that compare microbiome compositions between conditions. 
Our simulation studies demonstrated that the proposed MECAF test achieves desirable power than competing methods while having the type I error rate well controlled under various scenarios.
The usefulness of the proposed test is further illustrated with two real microbiome data analyses. The source code of the proposed method is freely available at https://github.com/JiyuanHu.

Conclusions: MECAF is a flexible differential abundance test and achieves statistical efficiency in analyzing high-throughput  microbiome data. The proposed new method will allow us to efficiently discover shifts of microbiome abundances between disease and treatment conditions, broadening our understanding of the disease and ultimately improving clinical diagnosis and treatment.