BEEM-Static: Accurate inference of ecological interactions from cross-sectional microbiome data

Abstract

The structure and function of diverse microbial communities is underpinned by ecological interactions that remain uncharacterized. With rapid adoption of next-generation sequencing for studying microbiomes, data-driven inference of microbial interactions based on abundance correlations is widely used, but with the drawback that ecological interpretations may not be possible. Leveraging cross-sectional microbiome datasets for unravelling ecological structure in a scalable manner thus remains an open problem. We present an expectation-maximization algorithm (BEEM-Static) that can be applied to cross-sectional datasets to infer interaction networks based on an ecological model (generalized Lotka-Volterra). The method exhibits robustness to violations in model assumptions by using statistical filters to identify and remove corresponding samples. Benchmarking against 10 state-of-the-art correlation based methods showed that BEEM-Static can infer presence and directionality of ecological interactions even with relative abundance data (AUC-ROC>0.85), a task that other methods struggle with (AUC-ROC<0.63). In addition, BEEM-Static can tolerate a high fraction of samples (up to 40%) being not at steady state or coming from an alternate model. Applying BEEM-Static to a large public dataset of human gut microbiomes (n = 4,617) identified multiple stable equilibria that better reflect ecological enterotypes with distinct carrying capacities and interactions for key species. Conclusion BEEM-Static provides new opportunities for mining ecologically interpretable interactions and systems insights from the growing corpus of microbiome data.

A guide for comparing microbial co‐occurrence networks

“…In this section, several network construction approaches were introduced step by step with the stool_met data set. For other network construction approaches not shown in this protocol (e.g., SpiecEasi [18], FlashWeave [19], and BEEM‐static [21]), please see the help document of the cal_network function of trans_network class in the microeco package.

…”

“…The studies on network comparisons [ 20 ] and approach reviews [ 10 ] have thoroughly discussed the robustness of different approaches and particularly recommended suitable network approaches depending upon different challenges. In addition, BEEM‐static method [ 21 ] is dedicated to seek out the interactions for cross‐sectional microbiome data with the generalized Lotka‐Volterra (gLV) model and an expectation‐maximization algorithm, offering a directed network to gain insight into microbial co‐occurrence in communities.…”

“…Broadly speaking, the interpretations on how network edges represent have been largely overlooked mainly due to its difficulty. Although recently developed approaches are dedicated to reveal the direct interactions among microbes [ 19 , 21 ], the applicability of these methods is still not clear when heterogeneous data sets are applied. So it is not recommended to explain co‐occurrence network as the interaction network [ 13 , 26 ], producing a dilemma for users on how to explain the co‐occurrence network and to what extent the co‐occurrence network can represent the species interactions.…”

Current Viewpoint on Female Urogenital Microbiome—The Cause or the Consequence?

“…However, it requires massive datasets (as the number of parameters grows quadratically with the number of species) and dense longitudinal sampling, which limits the wider application of such methods [69]. With the lack of longitudinal study data due to limited research budgets, ethical concerns and the aforementioned challenges, generalized Lotka-Volterra models can also be fitted to cross-sectional data from multiple communities by imposing some additional assumptions [70,71]. Nevertheless, such methods still suffer from statistical and biological limitations.…”