The role of gut microbiome in predicting diet response and developing personalized dietary recommendations has been increasingly recognized. Yet, we still lack comprehensive, genome-based insights into which gut microbes metabolize specific dietary compounds. Here, we leveraged the metabolic networks constructed from well-annotated microbial genomes to characterize the potential interactions between microbes and metabolites, specifically emphasizing the interactions between microbes and dietary compounds. We revealed a substantial, approximately fourfold variation in both the number of metabolites and dietary compounds in the microbial genome-scale metabolic networks across different genera, whereas species within the same genus showed a high metabolic similarity (mean coefficient of variation in microbial network degree CV = 0.023 for metabolites and 0.015 for dietary compounds). We found that the number of species that can utilize a metabolite drastically varies, ranging from 1 to 818 species, with some metabolites being used by a wide range of species (211 out of 1390 metabolites used by more than 95

Urinary tract infections (UTIs) are common infections that pose a critical burden on healthcare and society. Despite growing recognition that the human urinary tract harbors its own microbiome, its composition, functional potential, and alterations in UTI remain limited. Here, we leveraged the publicly available whole-metagenome shotgun sequencing data from 450 urinary microbiome samples collected in four independent cohorts together with genome assembly and metagenomic binning to construct an extensive human urinary microbiome catalog consisting of ∼1.3 million non-redundant microbial genes and 705 non-redundant metagenome-assembled genomes (nrMAGs). We found that microbiomes from patients with UTI carry significantly more genes linked to antibiotic resistance and virulence vs controls. There was an enrichment of multiple Escherichia strains in patients with UTI from two independent case-control cohorts. UTIs are becoming multidrug-resistant, and we used machine learning models to identify potential antimicrobial peptides (AMPs) in 705 nrMAGs. Furthermore, we experimentally demonstrated that two of these AMPs exhibited strong inhibitory activity against uropathogenic Escherichia coli strains. Our study provides a valuable resource for studying the human urinary microbiome and suggests urinary microbiome-derived AMPs represent a source of new therapeutics for UTIs.

Targeted perturbations of individual microbial taxa can propagate through complex ecological networks and generate ripple effects that reshape gut microbiota structure and function. Here, we discuss the need for predictive ecological and data-driven frameworks that enable precise and controllable microbiome engineering to minimize or leverage ripple effects.

Advancements in artificial intelligence (AI) have transformed many scientific fields, with microbiology and microbiome research now experiencing significant breakthroughs through machine-learning applications. This review provides a comprehensive overview of AI-driven approaches tailored for microbiology and microbiome studies, emphasizing both technical advancements and biological insights. We first introduce foundational AI techniques and offer guidance on choosing between traditional machine-learning and sophisticated deep-learning methods based on specific research goals. The primary section on application scenarios spans diverse research areas from taxonomic profiling, functional annotation and prediction, microbe-X interactions, microbial ecology, metabolic modeling, precision nutrition, and clinical microbiology to prevention and therapeutics. Finally, we discuss challenges in this field and highlight some recent breakthroughs. Together, this review underscores AI's transformative role in microbiology and microbiome research, paving the way for innovative methodologies and applications that enhance our understanding of microbial life and its impact on our planet and our health.

ABSTRACT The gut virome represents a vast reservoir of genetic diversity with profound implications for human health, yet it remains the “dark matter” of the microbiome due to the staggering complexity of reproducible viral profiling. It remains fundamentally contested whether biologically informative virome signals can be robustly recovered from routine whole-metagenome sequencing (WMS), and to what extent these signals offer ecological insights independent of the bacteriome. Here we present VIP2B, a framework that leverages Type IIB restriction tags to extract multifaceted viral features (encompassing taxonomy, coverage, function, and phenotype) directly from bulk WMS data. Through extensive benchmarking across incomplete references, unseen genomes, and high bacterial or host background, we demonstrate that VIP2B achieved high precision and robust taxonomic concordance. By applying VIP2B to paired bulk and virus-like particle (VLP)-enriched datasets, we reveal a species-level overlap far greater than previously recognized, proving that standard bulk metagenomes contain a wealth of recoverable viral information. Analysis of 20 clinical cohorts demonstrates that coverage-, function-, and phenotype-resolved viral features consistently identify disease-associated signatures that escape taxonomic analysis alone, significantly improving diagnostic models over bacteriome-only approaches. Finally, we define two distinct gut virome community states at the population scale (n=6,090), characterized by divergent diversity profiles and health associations. Our findings establish the gut virome as a non-redundant, clinically actionable component of the human holobiont and provide the methodology necessary to transition microbiome research toward a truly multi-kingdom framework.

BACKGROUND:The rising global health crisis of childhood overweight and obesity is potentially influenced by caesarean delivery (CD), but it remains a subject of ongoing debate. The gut microbiome, which is affected by delivery mode and can impact body weight, might play a role in this issue. However, the complex relationship between them remains poorly understood. METHODS:We analysed data from a randomised, double-blind, placebo-controlled trial VDAART cohort, including BMI percentiles from 683 children aged 2-8 years and 1672 stool samples collected between 3 months and 5 years (all data in this study were collected between May 2010 and February 2018). To evaluate how CD relates to BMI trajectories, we conducted permutation testing and discussed the effect of confounding factors. We then used PERMANOVA, random forest classification, and Generalised Microbe Phenotype Triangulation (GMPT) to explore the role of the gut microbiota in mediating this relationship. FINDINGS:Compared with vaginal delivery, intrapartum CD (iCD) rather than antepartum CD (aCD) was associated with a higher BMI percentile trajectory (Δ = 31.8%; 95% CI, 16.25%-47.55%; P = 0.001, Permutation test), and this was observed only among female children. Moreover, delivery mode was significantly associated with early-life gut microbiota, with effects also limited to females (F = 2.15 and 2.47 at months 3-6 and at age 1; P = 0.035 and 0.007, PERMANOVA). Random Forest models using early microbiota data can predict later overweight/obesity, performing best among iCD-born females (AUROC = 0.88; 95% CI, 0.83-0.94 for age 2), indicating an optimal intervention window before age one. Finally, GMPT identified 24 early-life taxa potentially mediating iCD-related overweight/obesity risk (11 preventive; 13 permissive), including Bacteroides ovatus, Bifidobacterium bifidum, Clostridium leptum, Eggerthella lenta, etc. INTERPRETATION: Our results indicate that CD types and children's sex are key factors in this interaction, offering a possible explanation for the ongoing debate about whether CD is linked to childhood overweight/obesity, and providing valuable insights for future intervention strategies. FUNDING:This work was supported by the National Institutes of Health.

Network dismantling aims to identify a set of critical nodes whose removal rapidly fragments a network's connectivity and functionality, with applications in controlling epidemics, halting rumor spread, and disrupting criminal networks. While previous studies have mainly focused on undirected networks, many real-world systems are directed, such as the World Wide Web and global trade networks. In directed networks, the giant strongly connected component captures mutual reachability and enables feedback loops that sustain system functionality. Here we introduce a centrality measure called network incoherence centrality and develop a trophic analysis-based dismantling method in which nodes are removed in descending order of their scores. Tested on synthetic networks and 14 real-world directed networks, our method consistently outperforms existing approaches. It also triggers the largest connectivity avalanches, highlighting its ability to pinpoint structurally critical nodes. These findings advance understanding of structure-function relationships in directed networks and inform the design of more resilient systems.

Abstract Background: Endogenous estrogens are well-established risk factors for breast cancer. A subset of gut microbes, collectively termed the estrobolome, plays a fundamental role in estrogen metabolism and reabsorption. However, existing studies examining associations between the gut microbiome and circulating estrogen levels in humans have been few and small in scale, and no prior work has established an estrobolome linked to circulating estrogen levels using shotgun metagenomic sequencing. Methods: This study included 141 non-hormone users from a microbiome sub-study nested within the Nurses’ Health Study II. Stool samples were collected using a home-based self-collection protocol, followed by blood collection approximately 3-4 months later. Circulating estrone (E1), estradiol (E2), and testosterone levels were assayed at the Mayo Clinic Laboratory by liquid chromatography-tandem mass spectrometry. Shotgun metagenomic sequencing was conducted using the 100 nt Illumina HiSeq platform. Taxonomic and functional profiling were performed using the bioBakery 4.0 workflow. Microbial α-diversity (Inverse Simpson index) and β-diversity (Bray-Curtis dissimilarity) were calculated. LASSO regression was applied to identify microbial features predictive of hormone levels, and age and BMI-adjusted generalized linear models were used for association analyses with multiple comparisons correction. Results: Taxonomic and functional profiling identified 1,860 species and 512 MetaCyc pathways. Microbial α-diversity was inversely correlated with E1 (R = -0.16, p = 0.07) and E2 (R = -0.16, p = 0.06), and positively correlated with testosterone (R = 0.12, p = 0.16). In age-adjusted models, α-diversity was significantly inversely associated with E2 (β = -0.39; 95% CI: -0.70, -0.07), though the association attenuated after further adjustment for BMI at stool collection (β = -0.16; 95% CI: -0.47, 0.15). PERMANOVA analyses revealed significant associations between taxonomic variation and E1 (R2 = 0.28%), E2 (R2 = 0.34%), and testosterone (R2 = 0.16%) (all p < 0.005). LASSO feature selection identified 23 species predictive of E1 and 16 predictive of E2, with the majority belonging to the genera Bacteroides, Clostridium, Eubacterium, and Alistipes. E2 was further associated with microbial pathways related to glycogen degradation and queuosine biosynthesis. Conclusion: This study represents the first and largest shotgun metagenomic investigation of gut microbiome composition in relation to circulating sex hormones in an epidemiological cohort. Reduced microbial diversity and higher abundance of estrobolome species with β-glucuronidase activity (e.g., Bacteroides fragilis) may be associated with elevated circulating estrogen levels, suggesting that structural variation in gut microbial communities contributes to inter-individual differences in estrogen metabolism. Citation Format: Tengteng Wang, Tong Cheng, Walter C. Willett, Curtis Huttenhower, Yang-Yu Liu, A. Heather Eliassen. Gut microbiome and breast cancer-related endogenous hormone risk factors in US women: findings from the Nurses’ Health Study II [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 2328.

Dietary intervention is an effective way to alter the gut microbiome to promote human health. Yet, due to our limited knowledge of diet-microbe interactions and the highly personalized gut microbial compositions, an efficient method to prescribe personalized dietary recommendations to achieve desired gut microbial compositions is still lacking. Here, we propose a machine learning framework to resolve this challenge. Our key idea is to implicitly learn the diet-microbe interactions by training a machine learning model using paired gut microbiome and dietary intake data from a population-level cohort. The well-trained machine learning model enables us to predict the microbial composition of any given species collection and dietary intake. Next, we prescribe personalized dietary recommendations by solving an optimization problem to achieve the desired microbial compositions. We systematically validated this Machine learning-based Personalized Dietary Recommendation (MPDR) framework using synthetic data generated from an established microbial consumer-resource model. We then validated MPDR using real data collected from a diet-microbiome association study. The presented MPDR framework demonstrates the potential of machine learning for personalized nutrition.

Due to highly personalized biological and lifestyle characteristics, different individuals may have different metabolite responses to specific foods and nutrients. In particular, the gut microbiota, a collection of trillions of microorganisms living in the gastrointestinal tract, is highly personalized and plays a key role in the metabolite responses to foods and nutrients. Accurately predicting metabolite responses to dietary interventions based on individuals’ gut microbial compositions holds great promise for precision nutrition. Existing prediction methods are typically limited to traditional machine learning models. Deep learning methods dedicated to such tasks are still lacking. Here we develop a method McMLP (Metabolite response predictor using coupled Multilayer Perceptrons) to fill in this gap. We provide clear evidence that McMLP outperforms existing methods on both synthetic data generated by the microbial consumer-resource model and real data obtained from six dietary intervention studies. Furthermore, we perform sensitivity analysis of McMLP to infer the tripartite food-microbe-metabolite interactions, which are then validated using the ground-truth (or literature evidence) for synthetic (or real) data, respectively. The presented tool has the potential to inform the design of microbiota-based personalized dietary strategies to achieve precision nutrition. Precision nutrition requires accurate predictions of individual metabolic responses to diets. Here, authors show their deep-learning model, McMLP, outperforms existing methods in predicting metabolite responses to dietary interventions.

In the developing neocortex, a diverse array of neurons with defined types and abundances are systematically generated by a limited population of radial glial progenitors (RGPs) as they undergo successive fate changes. The molecular regulation behind this intricate temporal patterning remains elusive. We undertook in-depth single-cell multi-omics analyses, discovering a two-layered regulatory framework at the core of this process. Central to this are global temporal regulators positioned above a temporal network, consisting of series of transcriptional factor (TF) hub groups under sequential state transitions. This temporal network operates not by restricting TF expressions to discrete temporal windows, but through coordinated transcriptional and chromatin-accessibility dynamics that modulate transient TF regulatory activity. Moreover, global temporal regulators specify the duration of each cascading stage and, consequently, the number of progenies generated at each stage. Loss of global temporal regulators protracts RGP lineage progression, whereas their increased activity accelerates it. These findings suggest a two-layer temporal regulatory system controlling RGP lineage progression and neural progeny output duality in mammalian neocortical development. ### Competing Interest Statement The authors have declared no competing interest.

Clostridioides difficile infection (CDI) is a major cause of healthcare- and antibiotic-associated diarrhea. While fecal microbiota transplantation (FMT) shows promise for recurrent CDI, its mechanisms and long-term safety are not fully understood. Live biotherapeutic products (LBPs) using predefined bacterial consortia offer an alternative option, but the rational design of LBPs remains challenging. Here, we employ a computational pipeline and three metagenomic datasets to identify microbial strains for LBPs targeting CDI. We constructed the CDI-related microbial genome catalog, comprising 3741 nonredundant metagenome-assembled genomes (nrMAGs), and identified multiple potential protective nrMAGs, including strains from Dorea formicigenerans, Oscillibacter welbionis, and Faecalibacterium prausnitzii. Importantly, some of these protective nrMAGs were found to play an important role in FMT success, and most top protective nrMAGs can be validated by various previous findings. Our results demonstrate a framework for selecting microbial strains targeting CDI, paving the way for the computational design of LBPs against other enteric infections.

Microbial communities are essential for sustaining ecosystem functions in diverse environments, including the human gut. Phages interact dynamically with their prokaryotic hosts and play a crucial role in shaping the structure and function of microbial communities. Previous approaches for inferring phage-host interactions (PHIs) from metagenomic data are constrained by low sensitivity and the inability to accurately capture ecological relationships. To overcome these limitations, we developed PHILM (Phage-Host Interaction Learning from Metagenomic profiles), a deep learning framework that predicts PHIs directly from the taxonomic profiles of metagenomic data. We validated PHILM on both synthetic datasets generated by ecological models and real-world data, finding that it consistently outperformed the co-abundance-based approach for inferring PHIs. When applied to a large-scale metagenomic dataset comprising 7,016 stool samples from healthy individuals, PHILM identified 90% more genus-level PHIs than the traditional assembly-based approach. In a longitudinal dataset tracking PHI dynamics, PHILM's latent representations recapitulated microbial succession patterns originally described using taxonomic abundances. Furthermore, we demonstrated that PHILM's latent representations served as more discriminative features than taxonomic abundance-based features for disease classifications. In summary, PHILM represents a novel computational framework for predicting phage-host interactions from metagenomic data, offering valuable insights for both microbiome science and translational medicine.

Genes associated with the same disease frequently engage in mutual biological interactions, e.g., perturbation within a specific neighborhood in the molecular interactome, often referred to as the disease module. This has propelled the advancement of network-based approaches toward elucidating the molecular bases of human diseases. Although many computational methods have been developed to integrate the molecular interactome and omics profiles to extract such context-dependent disease modules, approaches that leverage multi-omics for disease-module detection are still lacking. Here, we developed a statistical physics approach based on the random-field O(n) model (RFOnM) to fill this gap. We applied the RFOnM approach to integrate gene-expression data and genome-wide association studies or mRNA data and DNA methylation for several complex diseases with the human interactome. We found that the RFOnM approach outperforms existing single omics methods in most of the complex diseases considered in this study.

This study monitored gut microbiome changes in healthy volunteers following inulin intervention, revealing dynamic and highly individualized shifts in microbial composition and short-chain fatty acid production. Using in vitro batch cultures, correlation analysis, and predictive modeling, we explored the personalized microbiome response. Our findings highlight the individualized response of the gut microbiome to prebiotics and the need for precision nutrition.

Metastatic breast cancer (MBC) is a devastating disease, and recent evidence suggests that the human microbiome may play a critical role in cancer metastasis. However, the role of the blood microbiome in MBC prognosis has not been explored. This study analyzed sixteen blood samples collected from women diagnosed with MBC who participated in the hospital-based cohort at Thomas Jefferson University. Plasma DNA samples were extracted from these samples collected before initiating new lines of therapy. We utilized 2bRAD sequencing for Microbiome (2bRAD-M), a reduced sequencing technology capable of decoding the blood microbiome (bacteria, fungi, archaea, and viruses) with high precision and resolution. Circulating tumor cells (CTCs) were enriched from blood samples using CellSearch®. To investigate blood microbiome differences by CTC levels and race, both conventional statistical analysis and machine learning are performed to examine the variations in blood microbial diversity, taxonomic structures, and functional pathways. Cox proportional hazard regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for associations between microbial diversity index and MBC overall survival. We identified 265 microbial species in the plasma samples. Lower blood microbial alpha diversity was significantly correlated with higher CTC counts (R = -0.66, p = 0.01), a marker associated with poorer breast cancer outcomes. Random forest analysis identified fifteen key species (e.g., Sphingopyxis_sp001468225, Malassezia restricta) that can differentiate between CTC groups, with an area under the curve of 0.67 to 0.75. Pathway analysis revealed the bacterial Type VI Secretion System (T6SS) as the top pathway associated with differential microbes across CTC levels; women with a CTC count >5 showed over 1000-fold higher abundance of the T6SS pathway. For survival prediction, patients with low microbial alpha diversity (≤ median of 2.04) exhibited a nearly threefold increased risk of death (HR = 2.62, 95% CI=0.74 -17.26), with even higher risk after adjusting for tumor ER status (HR = 4.43), age at diagnosis (HR = 4.86), and CTC count (HR = 6.83). These findings indicate a strong negative association between blood microbial diversity and key prognostic factors in MBC. Lastly, we also observed that Black MBC patients had a lower microbial richness than White participants (median 27 vs. 42), and had a significantly lower abundance of Klebsiella pneumoniae than White women (p<0.001). Our compelling pilot findings suggest that 1) blood samples may serve as a novel sample type for microbiome assessment in a diverse cohort of women with MBC, and 2) blood microbiome features differ by CTC levels and race and may significantly influence MBC survival. Further studies with larger cohorts are warranted as the next step. Tengteng Wang, Prachi Trevedi, Mouadh Barbirou, Maysa Abu-Khalaf, Zheng Sun, Yang-Yu Liu, Mridula George, Elisa V. Bandera, Hushan Yang. Blood microbiome, circulating tumor cells, and survival in Black and White women with metastatic breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 2300.