Statistical analysis of microbiome data: The challenge of sparsity. (August 2021)

• Record Type: Journal Article
• Authors: Pan, Amy Y.
• Is Part Of: Current opinion in endocrine and metabolic research. Volume 19(2021)
• Page Start: 35
• Page End: 40
• Publication Date: 2021-08
• Subjects: Microbiome -- Sparsity -- Statistical analysis -- Zero-inflated models; Endocrine glands -- Diseases -- Periodicals; Metabolism -- Disorders -- Periodicals; Endocrine System Diseases; Metabolic Diseases; Endocrine glands -- Diseases; Metabolism -- Disorders; Periodical; Electronic journals; Periodicals; Electronic journals; 616.4
• DOI: 10.1016/j.coemr.2021.05.005
• Languages: English
• ISSNs: 2451-9650

Abstract

Abstract: Microbiomes not only exist across many different body sites in human beings but also interact dynamically with the host and environment. The unique feature and complexity of 16S ribosomal RNA gene sequence data, especially the sparsity of the data, present challenges to statistical analysis and interpretation. Proper normalization is critical in ensuring the validity of downstream analysis. Modeling the inflated zeros is an active area of research, and it has been extended from differential abundance analysis to a variety of other types of analyses. One specific method may not work well under all conditions, where sample size, effect size, data distribution, and degree of zero inflation vary greatly.