|
Abstract:
|
Background and objective: The gut human microbiome contains more cells than those present in the rest of our body. Advances in the human microbiome field have enabled us to use molecular and statistical techniques to investigate the influences of early life exposures on infant gut microbial composition and future disease risk. Our goal is to identify strengths and weaknesses of various methods proposed in recent years, and compare their performance by simulating composition human microbiome data. Method: To achieve the study goal, we will simulate microbial compositional data similar to that of the infant gut, employing published statistical methods in the human gut microbiome literature. Result: Based on the simulation study, we anticipate that existing methods will differ in their performance when dealing with microbial compositional data common problems, such as overdispersion, zero inflation and correlation. Conclusion: Microbial compositional data have unique characteristics which makes their analysis challenging. It is important to choose the appropriate statistical tests and methods according to proposed hypotheses and the underlying structure of this complex data.
|
