A1022
Title: Association test for longitudinal microbiome data
Authors: Taesung Park - Seoul National University (Korea, South) [presenting]
Nayeon Kang - Seoul National University (Korea, South)
Hyunwook Koh - The State University of New York, Korea (Korea, South)
Abstract: High-throughput technologies allow a new era of metagenomics studies to explore microbial communities sampled directly. The main goal of human microbial studies is to detect associations between microbiota and subject grouping phenotype. However, the microbiome data has several issues to overcome such as count compositional structure, various total sequence reads per sample, over-dispersion and zero-inflation. Several tools have been developed to handle these characteristics. We propose a permutation approach to identifying differentially abundant markers between two groups. The proposed method is based on the logistic regression model and has the advantage of handling multiple markers easily. Compared to existing methods, the proposed approach shows better performance in empirical studies including simulations and real data studies.
