EMBED: Essential Microbiome Dynamics, a dimensionality reduction approach for longitudinal microbiome studies

Abstract

The gut microbiome is well-established to be a significant driver of host health and disease. Longitudinal studies involving high-throughput sequencing technologies have begun to unravel the complex dynamics of these ecosystems, and quantitative frameworks are now being developed to better understand their organizing principles. Dimensionality reduction can offer unique insights into gut bacterial dynamics by leveraging collective abundance fluctuations of multiple bacteria driven by similar underlying ecological factors. However, methods providing lower-dimensional representations of gut microbial dynamics both at the community and individual taxa level are currently missing. To that end, we develop EMBED: Essential Microbiome Dynamics. Similar to normal modes in structural biology, EMBED infers ecological normal modes (ECNs), which represent the unique set of orthogonal dynamical trajectories capturing the collective behavior of a community. We show that a small number of ECNs accurately describe gut microbiome dynamics across data sets that encompass dietary changes and antibiotic-related perturbations. Importantly, we find that ECNs often reflect specific ecological behaviors, providing natural templates along which the dynamics of individual bacteria may be partitioned. Collectively, our results highlight the utility of dimensionality reduction approaches to understanding the dynamics of the gut microbiome and provide a framework to study the dynamics of other high-dimensional systems as well.

SMbiot: A Shared Latent Model for Microbiomes and their Hosts

“…(2), z ( P ) are collectively the latents that describe the metadata and Φ is the matrix of phenotype-related features. We model microbial abundances using a multinomial distribution, where is the reduced-dimensional description of the microbiome 17,18 . In Eq (4) z ( O ) are collectively the latents that describe the microbiome and Θ is the matrix of microbiome features.…”

“…is the reduced-dimensional description of the microbiome 17,18 . In Eq (4) ࢠ ሺைሻ are collectively the latents that describe the microbiome and દ is the matrix of microbiome features.…”

“…In Eq (4) Z (O) are collectively the latents that describe the microbiome and Θ is the matrix of microbiome features. We have previously shown that this Gibbs-Boltzmann form can approximate microbial abundances to a high degree of accuracy using a very small number of latents 1 .…”

“…We recently developed a computational method to embed temporal variability in gut microbiomes using latent space descriptors 17 . This allowed us to capture the collective variability in bacterial abundances in the gut microbiome using a small number of latent space dimensions.…”

SiGMoiD: A super-statistical generative model for binary data

“…We have previously developed SiGMoiD-like approaches [ 16 , 34 ] to model multinomially distributed abundance data common in sequencing studies including 16s sequencing based characterization of the microbiome [ 34 ]. Going forward, the most straightforward generalization to SiGMoiD is applying it to study amino acid/nucleotide variation in sequencing data.…”