A Guide to Dietary Pattern-Microbiome Data Integration

TaxaHFE: a machine learning approach to collapse microbiome datasets using taxonomic structure

Motivation

Biologists increasingly turn to machine learning models not just to predict, but to explain. Feature reduction is a common approach to improve both the performance and interpretability of models. However, some biological datasets, such as microbiome data, are inherently organized in a taxonomy, but these hierarchical relationships are not leveraged during feature reduction. We sought to design a feature engineering algorithm to exploit relationships in hierarchically organized biological data.

Results

We designed an algorithm, called TaxaHFE, to collapse information-poor features into their higher taxonomic levels. We applied TaxaHFE to six previously published datasets and found, on average, a 90% reduction in the number of features (SD = 5.1%) compared to using the most complete taxonomy. Using machine learning to compare the most resolved taxonomic level (i.e. species) against TaxaHFE-preprocessed features, models based on TaxaHFE features achieved an average increase of 3.47% in receiver operator curve area under the curve. Compared to other hierarchical feature engineering implementations, TaxaHFE introduces the novel ability to consider both categorical and continuous response variables to inform the feature set collapse. Importantly, we find TaxaHFE's ability to reduce hierarchically organized features to a more information-rich subset increases the interpretability of models.

Availability and implementation

TaxaHFE is available as a Docker image and as R code at https://github.com/aoliver44/taxaHFE.

Diversity of plant DNA in stool is linked to dietary quality, age, and household income

Eating a varied diet is a central tenet of good nutrition. Here, we develop a molecular tool to quantify human dietary plant diversity by applying DNA metabarcoding with the chloroplast trnL-P6 marker to 1,029 fecal samples from 324 participants across two interventional feeding studies and three observational cohorts. The number of plant taxa per sample (plant metabarcoding richness or pMR) correlated with recorded intakes in interventional diets and with indices calculated from a food frequency questionnaire in typical diets (ρ = 0.40 to 0.63). In adolescents unable to collect validated dietary survey data, trnL metabarcoding detected 111 plant taxa, with 86 consumed by more than one individual and four (wheat, chocolate, corn, and potato family) consumed by >70% of individuals. Adolescent pMR was associated with age and household income, replicating prior epidemiologic findings. Overall, trnL metabarcoding promises an objective and accurate measure of the number and types of plants consumed that is applicable to diverse human populations.

Diet Quality and the Fecal Microbiota in Adults in the American Gut Project

BACKGROUND

The Dietary Guidelines for Americans advises on dietary intake to meet nutritional needs, promote health, and prevent diseases. Diet affects the intestinal microbiota and is increasingly linked to health. It is vital to investigate the relationships between diet quality and the microbiota to better understand the impact of nutrition on human health.

OBJECTIVES

This study aimed to investigate the differences in fecal microbiota composition in adults from the American Gut Project based on their adherence to the Dietary Guidelines for Americans.

METHODS

This study was a cross-sectional analysis of the 16S sequencing and food frequency data of a subset of adults (n = 432; age = 18-60 y; 65% female, 89% white) participating in the crowdsourced American Gut Project. The Healthy Eating Index-2015 assessed the compliance with Dietary Guideline recommendations. The cohort was divided into tertiles based on Healthy Eating Index-2015 scores, and differences in taxonomic abundances and diversity were compared between high and low scorers.

RESULTS

The mean Total Score for low-scoring adults (58.1 ± 5.4) was comparable with the reported score of the average American adult (56.7). High scorers for the Total Score and components related to vegetables, grains, and dairy had greater alpha diversity than low scorers. High scorers in the fatty acid component had a lower alpha diversity than low scorers (95% CI: 0.35, 1.85). A positive log-fold difference in abundance of plant carbohydrate-metabolizing taxa in the families Lachnospiraceae and Ruminococcaceae was observed in high-scoring tertiles for Total Score, vegetable, fruit, and grain components (Benjamini-Hochberg; q < 0.05).

CONCLUSIONS

Adults with greater compliance to the Dietary Guidelines demonstrated higher diversity in their fecal microbiota and greater abundance of bacteria capable of metabolizing complex carbohydrates, providing evidence on how Dietary Guidelines support the gut microbiota.

Methods in Nutrition & Gut Microbiome Research: An American Society for Nutrition Satellite Session [13 October 2022]

The microbial cells colonizing the human body form an ecosystem that is integral to the regulation and maintenance of human health. Elucidation of specific associations between the human microbiome and health outcomes is facilitating the development of microbiome-targeted recommendations and treatments (e.g., fecal microbiota transplant; pre-, pro-, and post-biotics) to help prevent and treat disease. However, the potential of such recommendations and treatments to improve human health has yet to be fully realized. Technological advances have led to the development and proliferation of a wide range of tools and methods to collect, store, sequence, and analyze microbiome samples. However, differences in methodology at each step in these analytic processes can lead to variability in results due to the unique biases and limitations of each component. This technical variability hampers the detection and validation of associations with small to medium effect sizes. Therefore, the American Society for Nutrition (ASN) Nutritional Microbiology Group Engaging Members (GEM), sponsored by the Institute for the Advancement of Food and Nutrition Sciences (IAFNS), hosted a satellite session on methods in nutrition and gut microbiome research to review currently available methods for microbiome research, best practices, as well as tools and standards to aid in comparability of methods and results. This manuscript summarizes the topics and research discussed at the session. Consideration of the guidelines and principles reviewed in this session will increase the accuracy, precision, and comparability of microbiome research and ultimately the understanding of the associations between the human microbiome and health.

Microbiome epidemiology and association studies in human health

Studies of the human microbiome share both technical and conceptual similarities with genome-wide association studies and genetic epidemiology. However, the microbiome has many features that differ from genomes, such as its temporal and spatial variability, highly distinct genetic architecture and person-to-person variation. Moreover, there are various potential mechanisms by which distinct aspects of the human microbiome can relate to health outcomes. Recent advances, including next-generation sequencing and the proliferation of multi-omic data types, have enabled the exploration of the mechanisms that connect microbial communities to human health. Here, we review the ways in which features of the microbiome at various body sites can influence health outcomes, and we describe emerging opportunities and future directions for advanced microbiome epidemiology.

Unique Pakistani gut microbiota highlights population-specific microbiota signatures of type 2 diabetes mellitus

Biogeographic variations in the gut microbiota are pivotal to understanding the global pattern of host-microbiota interactions in prevalent lifestyle-related diseases. Pakistani adults, having an exceptionally high prevalence of type 2 diabetes mellitus (T2D), are one of the most understudied populations in microbiota research to date. The aim of the present study is to examine the gut microbiota across individuals from Pakistan and other populations of non-industrialized and industrialized lifestyles with a focus on T2D. The fecal samples from 94 urban-dwelling Pakistani adults with and without T2D were profiled by bacterial 16S ribosomal RNA gene and fungal internal transcribed spacer (ITS) region amplicon sequencing and eubacterial qPCR, and plasma samples quantified for circulating levels of lipopolysaccharide-binding protein (LBP) and the activation ability of Toll-like receptor (TLR)-signaling. Publicly available datasets generated with comparable molecular methods were retrieved for comparative analysis of the bacterial microbiota. Overall, urbanized Pakistanis' gut microbiota was similar to that of transitional or non-industrialized populations, depleted in Akkermansiaceae and enriched in Prevotellaceae (dominated by the non-Westernized clades of Prevotella copri). The relatively high proportion of Atopobiaceae appeared to be a unique characteristic of the Pakistani gut microbiota. The Pakistanis with T2D had elevated levels of LBP and TLR-signaling in circulation as well as gut microbial signatures atypical of other populations, e.g., increased relative abundance of Libanicoccus/Parolsenella, limiting the inter-population extrapolation of gut microbiota-based classifiers for T2D. Taken together, our findings call for a more global representation of understudied populations to extend the applicability of microbiota-based diagnostics and therapeutics.