A1319
Title: Advances in microbiome differential abundance analysis: Group-wise normalization and multivariate count regression
Authors: Kyu Ha Lee - Harvard T.H. Chan School of Public Health (United States) [presenting]
Abstract: The focus is on a new perspective on normalization for differential abundance analysis, highlighting a group-level approach to mitigate compositional bias. Within this framework, two novel strategies are introduced, group-wise relative log expression and fold-truncated sum scaling, that enhance statistical power while maintaining false discovery control. Then, recent developments in multivariate regression modeling for zero-inflated count data are discussed, with a particular focus on Bayesian variable selection. These models enable simultaneous inference across multiple taxa, scale effectively to high-dimensional datasets, automatically identify key associations, and integrate normalization procedures that address compositional challenges. Simulation studies and real-data applications illustrate that this combined framework surpasses existing univariate and multivariate methods, offering a robust and versatile toolkit for microbiome data analysis.
