Integrative analysis of relative abundance data and presence-absence data of the microbiome using the LDM

Evaluating changes in attractor sets under small network perturbations to infer reliable microbial interaction networks from abundance patterns

MOTIVATION

Inferring microbial interaction networks from microbiome data is a core task of computational ecology. An avenue of research to create reliable inference methods is based on a stylized view of microbiome data, starting from the assumption that the presences and absences of microbiomes, rather than the quantitative abundances, are informative about the underlying interaction network. With this starting point, inference algorithms can be based on the notion of attractors (asymptotic states) in Boolean networks. Boolean network framework offers a computationally efficient method to tackle this problem. However, often existing algorithms operating under a Boolean network assumption, fail to provide networks that can reproduce the complete set of initial attractors (abundance patterns). Therefore, there is a need for network inference algorithms capable of reproducing the initial stable states of the system.

RESULTS

We study the change of attractors in Boolean threshold dynamics on signed undirected graphs under small changes in network architecture and show, how to leverage these relationships to enhance network inference algorithms. As an illustration of this algorithmic approach, we analyse microbial abundance patterns from stool samples of humans with inflammatory bowel disease (IBD), with colorectal cancer and from healthy individuals to study differences between the interaction networks of the three conditions. The method reveals strong diversity in IBD interaction networks. The networks are first partially deduced by an earlier inference method called ESABO, then we apply the new algorithm developed here, EDAME, to this result to generate a network that comes nearest to satisfying the original attractors.

AVAILABILITY AND IMPLEMENTATION

Implementation code is freely available at https://github.com/Jojo6297/edame.git.

Association of Gut Microbiota With Fatigue in Black Women With Polycystic Ovary Syndrome

BACKGROUND

Fatigue is a highly prevalent symptom for individuals with polycystic ovary syndrome (PCOS); however, characterization of fatigue and investigation into the gut microbiome-a pathway that may contribute to fatigue-remains inadequately explored in Black women with PCOS.

OBJECTIVES

The purpose of this cross-sectional study was to examine fatigue and its relationship to the gut microbiome in adult Black women with PCOS.

METHODS

Adult Black women with a diagnosis of PCOS were recruited for this cross-sectional study. The Multidimensional Fatigue Inventory-20 (MFI-20) and the PROMIS Fatigue Short Form were used to measure fatigue. The V3/V4 region of the bacterial 16S rRNA gene was sequenced to investigate gut microbial composition. Relative abundance and diversity values were calculated.

RESULTS

We found that Black women with PCOS experience mild to moderate levels of fatigue. An inverse relationship between fatigue scores and alpha diversity values was found for the gut microbiome. We also found distinct beta diversity profiles based on fatigue. Lastly, when controlling for hypertension and body mass index, Ruminococcus bromii, Blautia obeum, Roseburia, and HT002 were associated with three subscales of the MFI-20.

DISCUSSION

Black women with PCOS experience mild to moderate fatigue. Clinicians should be cognizant of this population's increased risk for fatigue to adequately address their healthcare needs. We also found that gut microbial composition was associated with fatigue in Black women with PCOS. Specifically, a higher relative abundance of certain gut bacteria involved in short-chain fatty acid production and anti-inflammatory pathways was correlated with lower fatigue levels. Future studies should further investigate the link between the gut microbiome and fatigue to determine whether this relationship is causal as better insight could inform tailored diet and exercise interventions to alter the gut microbiome and reduce fatigue.

MaAsLin 3: Refining and extending generalized multivariable linear models for meta-omic association discovery

A key question in microbial community analysis is determining which microbial features are associated with community properties such as environmental or health phenotypes. This statistical task is impeded by characteristics of typical microbial community profiling technologies, including sparsity (which can be either technical or biological) and the compositionality imposed by most nucleotide sequencing approaches. Many models have been proposed that focus on how the relative abundance of a feature (e.g. taxon or pathway) relates to one or more covariates. Few of these, however, simultaneously control false discovery rates, achieve reasonable power, incorporate complex modeling terms such as random effects, and also permit assessment of prevalence (presence/absence) associations and absolute abundance associations (when appropriate measurements are available, e.g. qPCR or spike-ins). Here, we introduce MaAsLin 3 (Microbiome Multivariable Associations with Linear Models), a modeling framework that simultaneously identifies both abundance and prevalence relationships in microbiome studies with modern, potentially complex designs. MaAsLin 3 also newly accounts for compositionality with experimental (spike-ins and total microbial load estimation) or computational techniques, and it expands the space of biological hypotheses that can be tested with inference for new covariate types. On a variety of synthetic and real datasets, MaAsLin 3 outperformed current state-of-the-art differential abundance methods in testing and inferring associations from compositional data. When applied to the Inflammatory Bowel Disease Multi-omics Database, MaAsLin 3 corroborated many previously reported microbial associations with the inflammatory bowel diseases, but notably 77% of associations were with feature prevalence rather than abundance. In summary, MaAsLin 3 enables researchers to identify microbiome associations with higher accuracy and more specific association types, especially in complex datasets with multiple covariates and repeated measures.

What is the Difference between Conventional Drinking Water, Potable Reuse Water, and Nonpotable Reuse Water? A Microbiome Perspective

As water reuse applications expand, there is a need for more comprehensive means to assess water quality. Microbiome analysis could provide the ability to supplement fecal indicators and pathogen profiling toward defining a "healthy" drinking water microbiota while also providing insight into the impact of treatment and distribution. Here, we utilized 16S rRNA gene amplicon sequencing to identify signature features in the composition of microbiota across a wide spectrum of water types (potable conventional, potable reuse, and nonpotable reuse). A clear distinction was found in the composition of microbiota as a function of intended water use (e.g., potable vs nonpotable) across a very broad range of U.S. water systems at both the point of compliance (Betadisper p > 0.01; ANOSIM p < 0.01, r-stat = 0.71) and point of use (Betadisper p > 0.01; ANOSIM p < 0.01, r-stat = 0.41). Core and discriminatory analysis further served in identifying distinct differences between potable and nonpotable water microbiomes. Taxa were identified at both the phylum (Desulfobacterota, Patescibacteria, and Myxococcota) and genus (Aeromonas and NS11.12_marine_group) levels that effectively discriminated between potable and nonpotable waters, with the most discriminatory taxa being core/abundant in nonpotable waters (with few exceptions, such as Ralstonia being abundant in potable conventional waters). The approach and findings open the door to the possibility of microbial community signature profiling as a water quality monitoring approach for assessing efficacy of treatments and suitability of water for intended use/reuse application.

Education as Risk Factor of Mild Cognitive Impairment: The Link to the Gut Microbiome

BACKGROUND

With differences apparent in the gut microbiome in mild cognitive impairment (MCI) and dementia, and risk factors of dementia linked to alterations of the gut microbiome, the question remains if gut microbiome characteristics may mediate associations of education with MCI.

OBJECTIVES

We sought to examine potential mediation of the association of education and MCI by gut microbiome diversity or composition.

DESIGN

Cross-sectional study.

SETTING

Luxembourg, the Greater Region (surrounding areas in Belgium, France, Germany).

PARTICIPANTS

Control participants of the Luxembourg Parkinson's Study.

MEASUREMENTS

Gut microbiome composition, ascertained with 16S rRNA gene amplicon sequencing. Differential abundance, assessed across education groups (0-10, 11-16, 16+ years of education). Alpha diversity (Chao1, Shannon and inverse Simpson indices). Mediation analysis with effect decomposition was conducted with education as exposure, MCI as outcome and gut microbiome metrics as mediators.

RESULTS

After exclusion of participants below 50, or with missing data, n=258 participants (n=58 MCI) were included (M [SD] Age=64.6 [8.3] years). Higher education (16+ years) was associated with MCI (Odds ratio natural direct effect=0.35 [95% CI 0.15-0.81]. Streptococcus and Lachnospiraceae-UCG-001 genera were more abundant in higher education.

CONCLUSIONS

Education is associated with gut microbiome composition and MCI risk without clear evidence for mediation. However, our results suggest signatures of the gut microbiome that have been identified previously in AD and MCI to be reflected in lower education and suggest education as important covariate in microbiome studies.

Compositional analysis of microbiome data using the linear decomposition model (LDM)

SUMMARY

There are compelling reasons to test compositional hypotheses about microbiome data. We present here linear decomposition model-centered log ratio (LDM-clr), an extension of our LDM approach to allow fitting linear models to centered-log-ratio-transformed taxa count data. As LDM-clr is implemented within the existing LDM program, this extension enjoys all the features supported by LDM, including a compositional analysis of differential abundance at both the taxon and community levels, while allowing for a wide range of covariates and study designs for either association or mediation analysis.

AVAILABILITY AND IMPLEMENTATION

LDM-clr has been added to the R package LDM, which is available on GitHub at https://github.com/yijuanhu/LDM.

Comprehensive microbiome causal mediation analysis using MiMed on user-friendly web interfaces

It is a central goal of human microbiome studies to see the roles of the microbiome as a mediator that transmits environmental, behavioral, or medical exposures to health or disease outcomes. Yet, mediation analysis is not used as much as it should be. One reason is because of the lack of carefully planned routines, compilers, and automated computing systems for microbiome mediation analysis (MiMed) to perform a series of data processing, diversity calculation, data normalization, downstream data analysis, and visualizations. Many researchers in various disciplines (e.g. clinicians, public health practitioners, and biologists) are not also familiar with related statistical methods and programming languages on command-line interfaces. Thus, in this article, we introduce a web cloud computing platform, named as MiMed, that enables comprehensive MiMed on user-friendly web interfaces. The main features of MiMed are as follows. First, MiMed can survey the microbiome in various spheres (i) as a whole microbial ecosystem using different ecological measures (e.g. alpha- and beta-diversity indices) or (ii) as individual microbial taxa (e.g. phyla, classes, orders, families, genera, and species) using different data normalization methods. Second, MiMed enables covariate-adjusted analysis to control for potential confounding factors (e.g. age and gender), which is essential to enhance the causality of the results, especially for observational studies. Third, MiMed enables a breadth of statistical inferences in both mediation effect estimation and significance testing. Fourth, MiMed provides flexible and easy-to-use data processing and analytic modules and creates nice graphical representations. Finally, MiMed employs ChatGPT to search for what has been known about the microbial taxa that are found significantly as mediators using artificial intelligence technologies. For demonstration purposes, we applied MiMed to the study on the mediating roles of oral microbiome in subgingival niches between e-cigarette smoking and gingival inflammation. MiMed is freely available on our web server (http://mimed.micloud.kr).

Compositional analysis of microbiome data using the linear decomposition model (LDM)

Summary

There are compelling reasons to test compositional hypotheses about microbiome data. We present here LDM-clr, an extension of our linear decomposition model (LDM) approach to allow fitting linear models to centered-log-ratio-transformed taxa count data. As LDM-clr is implemented within the existing LDM program, it enjoys all the features supported by LDM, including a compositional analysis of differential abundance at both the taxon and community levels, while allowing for a wide range of covariates and study designs for either association or mediation analysis.

Availability and Implementation

LDM-clr has been added to the R package LDM, which is available on GitHub at https://github.com/yijuanhu/LDM .

Contact

yijuan.hu@emory.edu.

Supplementary information

Supplementary data are available at Bioinformatics online.

Testing microbiome associations with survival times at both the community and individual taxon levels

BACKGROUND

Finding microbiome associations with possibly censored survival times is an important problem, especially as specific taxa could serve as biomarkers for disease prognosis or as targets for therapeutic interventions. The two existing methods for survival outcomes, MiRKAT-S and OMiSA, are restricted to testing associations at the community level and do not provide results at the individual taxon level. An ad hoc approach testing each taxon with a survival outcome using the Cox proportional hazard model may not perform well in the microbiome setting with sparse count data and small sample sizes.

METHODS

We have previously developed the linear decomposition model (LDM) for testing continuous or discrete outcomes that unifies community-level and taxon-level tests into one framework. Here we extend the LDM to test survival outcomes. We propose to use the Martingale residuals or the deviance residuals obtained from the Cox model as continuous covariates in the LDM. We further construct tests that combine the results of analyzing each set of residuals separately. Finally, we extend PERMANOVA, the most commonly used distance-based method for testing community-level hypotheses, to handle survival outcomes in a similar manner.

RESULTS

Using simulated data, we showed that the LDM-based tests preserved the false discovery rate for testing individual taxa and had good sensitivity. The LDM-based community-level tests and PERMANOVA-based tests had comparable or better power than MiRKAT-S and OMiSA. An analysis of data on the association of the gut microbiome and the time to acute graft-versus-host disease revealed several dozen associated taxa that would not have been achievable by any community-level test, as well as improved community-level tests by the LDM and PERMANOVA over those obtained using MiRKAT-S and OMiSA.

CONCLUSIONS

Unlike existing methods, our new methods are capable of discovering individual taxa that are associated with survival times, which could be of important use in clinical settings.

Assembly of root-associated N2O-reducing communities of annual crops is governed by selection for nosZ clade I over clade II

The rhizosphere is a hotspot for denitrification. The nitrous oxide (N2O) reductase among denitrifiers and nondenitrifying N2O reducers is the only known N2O sink in the biosphere. We hypothesized that the composition of root-associated N2O-reducing communities when establishing on annual crops depend on soil type and plant species, but that assembly processes are independent of these factors and differ between nosZ clades I and II. Using a pot experiment with barley and sunflower and two soils, we analyzed the abundance, composition, and diversity of soil and root-associated N2O reducing communities by qPCR and amplicon sequencing of nosZ. Clade I was more abundant on roots compared to soil, while clade II showed the opposite. In barley, this pattern coincided with N2O availability, determined as potential N2O production rates, but for sunflower no N2O production was detected in the root compartment. Root and soil nosZ communities differed in composition and phylogeny-based community analyses indicated that assembly of root-associated N2O reducers was driven by the interaction between plant and soil type, with inferred competition being more influential than habitat selection. Selection between clades I and II in the root/soil interface is suggested, which may have functional consequences since most clade I microorganisms can produce N2O.

A rarefaction-without-resampling extension of PERMANOVA for testing presence-absence associations in the microbiome

MOTIVATION

PERMANOVA is currently the most commonly used method for testing community-level hypotheses about microbiome associations with covariates of interest. PERMANOVA can test for associations that result from changes in which taxa are present or absent by using the Jaccard or unweighted UniFrac distance. However, such presence-absence analyses face a unique challenge: confounding by library size (total sample read count), which occurs when library size is associated with covariates in the analysis. It is known that rarefaction (subsampling to a common library size) controls this bias but at the potential costs of information loss and the introduction of a stochastic component into the analysis.

RESULTS

Here, we develop a non-stochastic approach to PERMANOVA presence-absence analyses that aggregates information over all potential rarefaction replicates without actual resampling, when the Jaccard or unweighted UniFrac distance is used. We compare this new approach to three possible ways of aggregating PERMANOVA over multiple rarefactions obtained from resampling: averaging the distance matrix, averaging the (element-wise) squared distance matrix and averaging the F-statistic. Our simulations indicate that our non-stochastic approach is robust to confounding by library size and outperforms each of the stochastic resampling approaches. We also show that, when overdispersion is low, averaging the (element-wise) squared distance outperforms averaging the unsquared distance, currently implemented in the R package vegan. We illustrate our methods using an analysis of data on inflammatory bowel disease in which samples from case participants have systematically smaller library sizes than samples from control participants.

AVAILABILITY AND IMPLEMENTATION

We have implemented all the approaches described above, including the function for calculating the analytical average of the squared or unsquared distance matrix, in our R package LDM, which is available on GitHub at https://github.com/yijuanhu/LDM.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

A new approach to testing mediation of the microbiome at both the community and individual taxon levels

MOTIVATION

Understanding whether and which microbes played a mediating role between an exposure and a disease outcome are essential for researchers to develop clinical interventions to treat the disease by modulating the microbes. Existing methods for mediation analysis of the microbiome are often limited to a global test of community-level mediation or selection of mediating microbes without control of the false discovery rate (FDR). Further, while the null hypothesis of no mediation at each microbe is a composite null that consists of three types of null, most existing methods treat the microbes as if they were all under the same type of null, leading to excessive false positive results.

RESULTS

We propose a new approach based on inverse regression that regresses the microbiome data at each taxon on the exposure and the exposure-adjusted outcome. Then, the P-values for testing the coefficients are used to test mediation at both the community and individual taxon levels. This approach fits nicely into our Linear Decomposition Model (LDM) framework, so our new method LDM-med, implemented in the LDM framework, enjoys all the features of the LDM, e.g. allowing an arbitrary number of taxa to be tested simultaneously, supporting continuous, discrete, or multivariate exposures and outcomes (including survival outcomes), and so on. Using extensive simulations, we showed that LDM-med always preserved the FDR of testing individual taxa and had adequate sensitivity; LDM-med always controlled the type I error of the global test and had compelling power over existing methods. The flexibility of LDM-med for a variety of mediation analyses is illustrated by an application to a murine microbiome dataset, which identified several plausible mediating taxa.

AVAILABILITY AND IMPLEMENTATION

Our new method has been added to our R package LDM, which is available on GitHub at https://github.com/yijuanhu/LDM.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.