Machine learning approaches in microbiome research: challenges and best practices

Frontiers in Microbiology, Journal Year: 2023, Volume and Issue: 14

Published: Sept. 22, 2023

Microbiome data predictive analysis within a machine learning (ML) workflow presents numerous domain-specific challenges involving preprocessing, feature selection, modeling, performance estimation, model interpretation, and the extraction of biological information from results. To assist decision-making, we offer set recommendations on algorithm pipeline creation evaluation, stemming COST Action ML4Microbiome. We compared suggested approaches multi-cohort shotgun metagenomics dataset colorectal cancer patients, focusing their in disease diagnosis biomarker discovery. It is demonstrated that use compositional transformations filtering methods as part preprocessing does not always improve model. In contrast, multivariate such Statistically Equivalent Signatures algorithm, was effective reducing classification error. When validated separate test dataset, this combination with random forest provided most accurate estimates. Lastly, showed how linear modeling by logistic regression coupled visualization techniques Individual Conditional Expectation (ICE) plots can yield interpretable results insights. These findings are significant for clinicians non-experts alike translational applications.

Language: Английский

---

The gut virome is associated with stress-induced changes in behaviour and immune responses in mice

Nature Microbiology, Journal Year: 2024, Volume and Issue: 9(2), P. 359 - 376

Published: Feb. 5, 2024

The microbiota-gut-brain axis has been shown to play an important role in the stress response, but previous work focused primarily on of bacteriome. gut virome constitutes a major portion microbiome, with bacteriophages having potential remodel bacteriome structure and activity. Here we use mouse model chronic social stress, employ 16S rRNA whole metagenomic sequencing faecal pellets determine how is modulated by contributes effects stress. We found that led behavioural, immune alterations mice were associated changes bacteriophage class Caudoviricetes unassigned viral taxa. To whether these causally related stress-associated behavioural or physiological outcomes, conducted transplant from before autochthonously transferred it undergoing transfer protected against behaviour sequelae restored stress-induced select circulating cell populations, cytokine release, gene expression amygdala. These data provide evidence plays modulation during indicating populations should be considered when designing future microbiome-directed therapies.

Language: Английский

---

Explainable artificial intelligence for microbiome data analysis in colorectal cancer biomarker identification

Frontiers in Microbiology, Journal Year: 2024, Volume and Issue: 15

Published: Feb. 15, 2024

Colorectal cancer (CRC) is a type of tumor caused by the uncontrolled growth cells in mucosa lining last part intestine. Emerging evidence underscores an association between CRC and gut microbiome dysbiosis. The high mortality rate this has made it necessary to develop new early diagnostic methods. Machine learning (ML) techniques can represent solution evaluate interaction intestinal microbiota host physiology. Through explained artificial intelligence (XAI) possible individual contributions microbial taxonomic markers for each subject. Our work also implements Shapley Method Additive Explanations (SHAP) algorithm identify subject which parameters are important context CRC.

Language: Английский

---

Integration of 168,000 samples reveals global patterns of the human gut microbiome

Cell, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 1, 2025

The factors shaping human microbiome variation are a major focus of biomedical research. While other fields have used large sequencing compendia to extract insights requiring otherwise impractical sample sizes, the field has lacked comparably sized resource for 16S rRNA gene amplicon commonly quantify composition. To address this gap, we processed 168,464 publicly available gut samples with uniform pipeline. We use compendium evaluate geographic and technical effects on variation. find that regions such as Central Southern Asia differ significantly from more thoroughly characterized microbiomes Europe Northern America composition alone can be predict sample's region origin. also strong associations between primers DNA extraction. anticipate growing work, Human Microbiome Compendium, will enable advanced applied methodological

Language: Английский

---

YACHT: an ANI-based statistical test to detect microbial presence/absence in a metagenomic sample

Bioinformatics, Journal Year: 2024, Volume and Issue: 40(2)

Published: Jan. 24, 2024

Abstract Motivation In metagenomics, the study of environmentally associated microbial communities from their sampled DNA, one most fundamental computational tasks is that determining which genomes a reference database are present or absent in given sample metagenome. Existing tools generally return point estimates, with no confidence uncertainty it. This has led to practitioners experiencing difficulty when interpreting results these tools, particularly for low-abundance organisms as often reside “noisy tail” incorrect predictions. Furthermore, few account fact databases incomplete and rarely, if ever, contain exact replicas an derived Results We solutions issues by introducing algorithm YACHT: Yes/No Answers Community membership via Hypothesis Testing. approach introduces statistical framework accounts sequence divergence between genomes, terms ANI, well sequencing depth, thus providing hypothesis test presence absence genome sample. After our approach, we quantify its power how this changes varying parameters. Subsequently, perform extensive experiments using both simulated real data confirm accuracy scalability approach. Availability implementation The source code implementing available Conda at https://github.com/KoslickiLab/YACHT. also provide reproducing https://github.com/KoslickiLab/YACHT-reproducibles.

Language: Английский

---

Gastric Microbiota Associated with Gastric Precancerous Lesions in Helicobacter pylori-Negative Patients

Microorganisms, Journal Year: 2025, Volume and Issue: 13(1), P. 81 - 81

Published: Jan. 3, 2025

Studies on the gastric microbiota associated with precancerous lesions remain limited. This study aimed to profile mucosal in patients Helicobacter pylori-negative lesions. Gastric samples were obtained from 67 H. patients, including those chronic gastritis (CG), intestinal metaplasia (IM), and dysplasia. The V3–V4 region of 16S rRNA gene was sequenced analyzed. No significant difference observed alpha or beta diversity among groups. However, a taxonomic analysis revealed enrichment Lautropia mirabilis depletion Limosilactobacillus reuteri, Solobacxterium moorei, Haemophilus haemolyticus, Duncaniella dubosii IM dysplasia groups compared CG group. Prevotella jejuni genus Parvimonas enriched A predictive functional ornithine degradation pathway groups, suggesting its role persistent inflammation showed an increased abundance oral microbes linked cancer reduction anti-inflammatory bacteria. These alterations might contribute inflammation, promoting carcinogenesis absence pylori infection.

Language: Английский

---

A Generalized Bayesian Stochastic Block Model for Microbiome Community Detection

Statistics in Medicine, Journal Year: 2025, Volume and Issue: 44(3-4)

Published: Jan. 24, 2025

ABSTRACT Advances in next‐generation sequencing technology have enabled the high‐throughput profiling of metagenomes and accelerated microbiome studies. Recently, there has been a rise quantitative studies that aim to decipher co‐occurrence network its underlying community structure based on metagenomic sequence data. Uncovering complex is essential understanding role disease progression susceptibility. Taxonomic abundance data generated from technologies are high‐dimensional compositional, suffering uneven sampling depth, over‐dispersion, zero‐inflation. These characteristics often challenge reliability current methods for detection. To study perform detection, we propose generalized Bayesian stochastic block model tailored analysis where transformed using recently developed modified centered‐log ratio transformation. Our also allows us leverage taxonomic tree information Markov random field prior. The parameters jointly inferred by chain Monte Carlo techniques. simulation showed proposed approach performs better than competing even when non‐informative. We applied our real urinary dataset postmenopausal women. best knowledge, this first time women studied. In summary, statistical methodology provides new tool facilitating advanced

Language: Английский

---

Organic cropping systems alter metabolic potential and carbon, nitrogen and phosphorus cycling capacity of soil microbial communities

Soil Biology and Biochemistry, Journal Year: 2025, Volume and Issue: unknown, P. 109737 - 109737

Published: Feb. 1, 2025

Language: Английский

---

Tree mycorrhizal type and tree diversity shape the forest soil microbiota

Environmental Microbiology, Journal Year: 2021, Volume and Issue: 24(9), P. 4236 - 4255

Published: July 30, 2021

Summary There is limited knowledge on how the association of trees with different mycorrhizal types shapes soil microbial communities in context changing tree diversity levels. We used arbuscular (AM) and ectomycorrhizal (EcM) species as con‐ heterospecific pairs (TSPs), which were established plots three levels including monocultures, two‐species mixtures multi‐tree a experiment subtropical China. found that type had significant effect fungal but not bacterial alpha diversity. Furthermore, only EcM AM TSPs increased diversity, differences between disappeared multi‐species mixtures. Tree type, their interaction effects community composition. Neither fungi nor bacteria showed any compositional variation located Accordingly, most influential taxa driving at low Collectively, our results indicate an important factor determining composition microbes, higher promote convergence communities. Significance statement More than 90% terrestrial plants have symbiotic associations could influence coexisting microbiota. Systematic understanding individual interactive microbiota crucial for mechanistic comprehension role microbes forest ecological processes. Our pair (TSP) concept coupled random sampling within across plots, allowed us unbiased assessment tree‐tree zone Unlike monocultures mixtures, we identified species‐rich converging Consequently, recommend planting trees, afforestation reforestation regimes. Specifically, findings highlight significance studying ‘tree – ecosystem function’ relationships.

Language: Английский

---

Depression and fatigue in active IBD from a microbiome perspective—a Bayesian approach to faecal metagenomics

BMC Medicine, Journal Year: 2022, Volume and Issue: 20(1)

Published: Oct. 17, 2022

Abstract Background Extraintestinal symptoms are common in inflammatory bowel diseases (IBD) and include depression fatigue. These highly prevalent especially active disease, potentially due to inflammation-mediated changes the microbiota-gut-brain axis. The aim of this study was investigate associations between structural functional microbiota characteristics severity fatigue depressive patients with IBD. Methods We included clinical data 62 prospectively enrolled IBD an disease state. Patients supplied stool samples completed questionnaires regarding symptoms. Based on taxonomic metagenomic profiles faecal gut microbiota, we used Bayesian statistics associative networks triangle motifs bacterial genera, modules symptom self-reported depression. Results Associations moderate strong evidence were found for 3 genera ( Odoribacter , Anaerotruncus Alistipes ) (pectin, glycosaminoglycan central carbohydrate metabolism) regard 4 Intestinimonas Eubacterium Clostridiales g.i.s) 2 implicating amino acid metabolism Conclusions This provides first association triplets composition, function extraintestinal Depression associated lower abundances short-chain fatty producers distinct pathways glycan, metabolism. Our results suggest that microbiota-directed therapeutic approaches may reduce should be investigated future research.

Language: Английский

---