Janson, L. | Search bioRxiv

bioRxiv2024

Compositional Differential Abundance Testing: Defining and Finding a New Type of Health-Microbiome Associations

A major task of microbiome epidemiology is association analysis, where the goal is to identify microbial features related to host health. This is commonly performed by differential abundance (DA) analysis, which, by design, examines each microbe as isolated from the rest of the microbiome. This does not properly account for the microbiomes compositional nature or microbe-microbe ecological interactions, and can lead to confounded findings, i.e., microbes that only appear to associate with health through their confounding association with health-related, biologically informative microbes. To remedy these issues, we present Compositional Differential Abundance (CompDA) analysis, a novel approach for health-microbiome association. CompDA provides a novel approach to identify health-related microbes by examining the microbiome holistically, which a) accounts for the datas compositionality and ecological interactions, and b) has clear interpretations corresponding to host health as affected by microbiome-based interventions. CompDA prioritizes health-related microbes and controls false discoveries by implementing recent advances from high-dimensional statistics, and can be flexibly adapted to many common tasks in modern microbiome epidemiology, including enhancing microbiome-based machine learning by providing rigorous p-values to prioritize important features. We validate the performance of CompDA, and compare against canonical microbiome association methods including DA with extensive, real-data-informed simulation studies. Lastly, we report novel and consistent findings of CompDA in application, based on re-examination of recently reported microbial signatures of colorectal cancer in a meta-analysis.

genomics↗

bioRxiv2021

A Statistical Model for Describing and Simulating Microbial Community Profiles

Many methods have been developed for statistical analysis of microbial community profiles, but due to the complex nature of typical microbiome measurements (e.g. sparsity, zero-inflation, nonindependence, and compositionality) and of the associated underlying biology, it is difficult to compare or evaluate such methods within a single systematic framework. To address this challenge, we developed SparseDOSSA (Sparse Data Observations for the Simulation of Synthetic Abundances): a statistical model of microbial ecological population structure, which can be used to parameterize real-world microbial community profiles and to simulate new, realistic profiles of known structure for methods evaluation. Specifically, SparseDOSSAs model captures marginal microbial feature abundances as a zero-inflated log-normal distribution, with additional model components for absolute cell counts and the sequence read generation process, microbemicrobe, and microbe-environment interactions. Together, these allow fully known covariance structure between synthetic features (i.e. "taxa") or between features and "phenotypes" to be simulated for method benchmarking. Here, we demonstrate SparseDOSSAs performance for 1) accurately modeling human-associated microbial population profiles; 2) generating synthetic communities with controlled population and ecological structures; 3) spiking-in true positive synthetic associations to benchmark analysis methods; and 4) recapitulating an end-to-end mouse microbiome feeding experiment. Together, these represent the most common analysis types in assessment of real microbial community environmental and epidemiological statistics, thus demonstrating SparseDOSSAs utility as a general-purpose aid for modeling communities and evaluating quantitative methods. An open-source implementation is available at http://huttenhower.sph.harvard.edu/sparsedossa2.

bioinformatics↗