CoMMiT: Co-informed inference of microbiome-metabolome interactions via transfer learning
Authors:
Leiyue Li,
Chenglong Ye,
Tim Randolph,
Meredith Hullar,
Johanna Lampe,
Marian Neuhouser,
Daniel Raftery,
Yue Wang
Abstract:
Recent multi-omic microbiome studies enable integrative analysis of microbes and metabolites, uncovering their associations with various host conditions. Such analyses require multivariate models capable of accounting for the complex correlation structures between microbes and metabolites. However, existing multivariate models often suffer from low statistical power for detecting microbiome-metabo…
▽ More
Recent multi-omic microbiome studies enable integrative analysis of microbes and metabolites, uncovering their associations with various host conditions. Such analyses require multivariate models capable of accounting for the complex correlation structures between microbes and metabolites. However, existing multivariate models often suffer from low statistical power for detecting microbiome-metabolome interactions due to small sample sizes and weak biological signals. To address these challenges, we introduce CoMMiT, Co-informed inference of Microbiome-Metabolome Interactions via novel Transfer learning models. Unlike conventional transfer-learning methods that borrow information from external datasets, CoMMiT leverages similarities across metabolites within a single cohort, reducing the risk of negative transfer often caused by differences in sequencing platforms and bioinformatic pipelines across studies. CoMMiT operates under the flexible assumption that auxiliary metabolites are collectively informative for the target metabolite, without requiring individual auxiliary metabolites to be informative. CoMMiT uses a novel data-driven approach to selecting the optimal set of auxiliary metabolites. Using this optimal set, CoMMiT employs a de-biasing framework to enable efficient calculation of p-values, facilitating the identification of statistically significant microbiome-metabolome interactions. Applying CoMMiT to a feeding study reveals biologically meaningful microbiome-metabolome interactions under a low glycemic load diet, demonstrating the diet-host link through gut metabolism.
△ Less
Submitted 30 June, 2025;
originally announced June 2025.
Kernel-Penalized Regression for Analysis of Microbiome Data
Authors:
Timothy W. Randolph,
Sen Zhao,
Wade Copeland,
Meredith Hullar,
Ali Shojaie
Abstract:
The analysis of human microbiome data is often based on dimension-reduced graphical displays and clustering derived from vectors of microbial abundances in each sample. Common to these ordination methods is the use of biologically motivated definitions of similarity. Principal coordinate analysis, in particular, is often performed using ecologically defined distances, allowing analyses to incorpor…
▽ More
The analysis of human microbiome data is often based on dimension-reduced graphical displays and clustering derived from vectors of microbial abundances in each sample. Common to these ordination methods is the use of biologically motivated definitions of similarity. Principal coordinate analysis, in particular, is often performed using ecologically defined distances, allowing analyses to incorporate context-dependent, non-Euclidean structure. Here we describe how to take a step beyond ordination plots and incorporate this structure into high-dimensional penalized regression models. Within this framework, the estimate of a regression coefficient vector is obtained via the joint eigen properties of multiple similarity matrices, or kernels. This allows for multivariate regression models to incorporate both a matrix of microbial abundances and, for instance, a matrix of phylogenetically-informed similarities between the abundance profiles. Further, we show how this regression framework can be used to address the compositional nature of multivariate predictors comprised of relative abundances; that is, vectors whose entries sum to a constant. We illustrate this regression framework with several simulations using data from two recent studies on the gut and vaginal microbiome. We conclude with an application to our own data, where we also incorporate a significance test for the estimated coefficients that represent associations between microbial abundance and a response.
△ Less
Submitted 9 January, 2017; v1 submitted 1 November, 2015;
originally announced November 2015.
