|
Activity Number:
|
346
- Advances in Diagnostics and Reproducibility Research
|
|
Type:
|
Topic-Contributed
|
|
Date/Time:
|
Thursday, August 12, 2021 : 10:00 AM to 11:50 AM
|
|
Sponsor:
|
Social Statistics Section
|
|
Abstract #317244
|
|
|
Title:
|
Assessing the Reproducibility of Microbiome Measurements Based on Concordance Correlation Coefficients
|
|
Author(s):
|
Limin Peng* and Ying Cui and Yijuan Hu and HuiChuan Lai
|
|
Companies:
|
Emory University and Emory University and Emory University and University of Wisconsin-Madison
|
|
Keywords:
|
Agreement;
Concordance correlation coefficient;
Microbiome compositional data;
high-dimensional vector;
Iota coefficient
|
|
Abstract:
|
Evaluating the reproducibility or agreement of microbiome measurements is often a crucial step to ensure rigorous downstream analyses in microbiome studies. In this paper, we address this need by developing adaptations of Lin's concordance correlation coefficient (CCC) tailored to microbiome studies. We introduce a general formulation of the new CCC measures upon the use of a distance function appropriately characterizing the discrepancy between microbiome compositional measurements. We thoroughly study the special cases that adopt Euclidean distance and Aitchison distance. Our proposals appropriately account for the unique features of microbiome compositional data, including high-dimensionality, dependency among individual relative abundances, and the presence of many zeros. We further investigate a practical compound approach to help better understand the sources of data inconsistency.We present extensive simulation studies and an application to a motivating dataset from the Feeding Infants Right.. from the STart (FIRST) study. Our empirical results provide useful insight about the utility of the proposed agreement methods in practical settings.
|
Authors who are presenting talks have a * after their name.
