Capturing the dynamics of microbial interactions through individual-specific networks

A Scoping Review of Artificial Intelligence for Precision Nutrition

BACKGROUND

With the role of artificial intelligence (AI) in precision nutrition rapidly expanding, a scoping review on recent studies and potential future directions is demanded.

OBJECTIVE

This scoping review examines: (1) the current landscape, including publication venues, targeted diseases, AI applications, methods, evaluation metrics, and considerations of minority and cultural factors; (2) common patterns in AI-driven precision nutrition studies; and (3) gaps, challenges and future research directions.

METHODS

Following the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) process, we extracted 198 articles from major databases with search keywords in three categories: precision nutrition keywords, artificial intelligence keywords and natural language processing keywords.

RESULTS

The extracted literature reveals a surge in AI-driven precision nutrition research, with ∼75% (n=148) published since 2020. It also showcases a diverse publication landscape, with these studies predominantly focusing on diet-related diseases, such as diabetes and cardiovascular conditions, while emphasizing health optimization, disease prevention, and management. We highlight diverse datasets and critically discuss methodologies and evaluation metrics to guide future studies. Importantly, we underscore the significance of minority and cultural aspects in enhancing health technologies and advancing equity. Future research should deepen the integration of these factors to fully harness AI's potential in precision nutrition.

STATEMENT OF SIGNIFICANCE

This scoping review offers the most recent advancements in artificial intelligence for precision nutrition studies, expanding the scope to not only AI methodologies and their applications but also evaluates publication venues, targeted disease, datasets used and minority and cultural factors, which have been mostly overlooked in prior studies. Furthermore, with numerous gaps and challenges presented in the discussion section, this review significantly improves the understanding of AI's role in precision nutrition and provides new insights for future research.

Individual-network based predictions of microbial interaction signatures for response to biological therapies in IBD patients

Inflammatory Bowel Disease (IBD), which includes Ulcerative Colitis (UC) and Crohn's Disease (CD), is marked by dysbiosis of the gut microbiome. Despite therapeutic interventions with biological agents like Vedolizumab, Ustekinumab, and anti-TNF agents, the variability in clinical, histological, and molecular responses remains significant due to inter-individual and inter-population differences. This study introduces a novel approach using Individual Specific Networks (ISNs) derived from faecal microbial measurements of IBD patients across multiple cohorts. These ISNs, constructed from baseline and follow-up data post-treatment, successfully predict therapeutic outcomes based on endoscopic remission criteria. Our analysis revealed that ISNs characterised by core gut microbial families, including Lachnospiraceae and Ruminococcaceae, are predictive of treatment responses. We identified significant changes in abundance levels of specific bacterial genera in response to treatment, confirming the robustness of ISNs in capturing both linear and non-linear microbiota signals. Utilising network topological metrics, we further validated these findings, demonstrating that critical microbial features identified through ISNs can differentiate responders from non-responders with respect to various therapeutic outcomes. The study highlights the potential of ISNs to provide individualised insights into microbiota-driven therapeutic responses, emphasising the need for larger cohort studies to enhance the accuracy of molecular biomarkers. This innovative methodology paves the way for more personalised and effective treatment strategies in managing IBD.

Bridging the gap in precision medicine: TranSYS training programme for next-generation scientists

Introduction

In the evolving healthcare landscape, precision medicine's rise necessitates adaptable doctoral training. The European Union has recognized this and promotes the development of international, training-focused programmes called Innovative Training Networks (ITNs). In this article, we introduce TranSYS, an ITN focused on educating the next generation of precision medicine researchers. In an ambition to go beyond describing the consortium goals, our article explores two key aspects of ITNs: the training and collaboration.

Methods

Using self-report questionnaires, we evaluate the scientific, professional, and personal growth of ESRs over the duration of the ITN and investigate whether this can be linked to network activities.

Results

Our quantitative analysis approach reveals substantial improvements in scientific, professional, and social skills among young researchers facilitated by the engagement in this interdisciplinary network. We provide case studies underlining the advantages of collaborative environments, featuring innovative scientific exchange within TranSYS.

Discussion

While challenging, ITNs foster positive growth in young researchers, yet exhibit weaknesses such as balancing stakeholder interests and partner commitment. We believe this study may benefit a variety of stakeholders, from prospective ITN creators to industry partners, to design better sustainable training networks going forward.

Disentangling the effects of terroir, season, and vintage on the grapevine fungal pathobiome

The composition, diversity and dynamics of microbial communities associated with grapevines may be influenced by various environmental factors, including terroir, vintage, and season. Among these factors, terroir stands out as a unique possible determinant of the pathobiome, the community of plant-associated pathogens. This study employed high-throughput molecular techniques, including metabarcoding and network analysis, to investigate the compositional dynamics of grapevine fungal pathobiome across three microhabitats (soil, woody tissue, and bark) using the Furmint cultivar. Samples were collected during late winter and late summer in 2020 and 2021, across three distinct terroirs in Hungary's Tokaj wine region. Of the 123 plant pathogenic genera found, Diplodia, Phaeomoniella, and Fusarium displayed the highest richness in bark, wood, and soil, respectively. Both richness and abundance exhibited significant disparities across microhabitats, with plant pathogenic fungi known to cause grapevine trunk diseases (GTDs) demonstrating highest richness and abundance in wood and bark samples, and non-GTD pathogens prevailed soil. Abundance and richness, however, followed distinct patterns Terroir accounted for a substantial portion of the variance in fungal community composition, ranging from 14.46 to 24.67%. Season and vintage also contributed to the variation, explaining 1.84 to 2.98% and 3.67 to 6.39% of the variance, respectively. Notably, significant compositional differences in fungi between healthy and diseased grapevines were only identified in wood and bark samples. Cooccurrence networks analysis, using both unweighted and weighted metrics, revealed intricate relationships among pathogenic fungal genera. This involved mostly positive associations, potentially suggesting synergism, and a few negative relationships, potentially suggesting antagonistic interactions. In essence, the observed differences among terroirs may stem from environmental filtering due to varied edaphic and mesoclimatic conditions. Temporal weather and vine management practices could explain seasonal and vintage fungal dynamics. This study provides insights into the compositional dynamics of grapevine fungal pathobiome across different microhabitats, terroirs, seasons, and health statuses. The findings emphasize the importance of considering network-based approaches in studying microbial communities and have implications for developing improved viticultural plant health strategies.

netMUG: a novel network-guided multi-view clustering workflow for dissecting genetic and facial heterogeneity

Introduction: Multi-view data offer advantages over single-view data for characterizing individuals, which is crucial in precision medicine toward personalized prevention, diagnosis, or treatment follow-up. Methods: Here, we develop a network-guided multi-view clustering framework named netMUG to identify actionable subgroups of individuals. This pipeline first adopts sparse multiple canonical correlation analysis to select multi-view features possibly informed by extraneous data, which are then used to construct individual-specific networks (ISNs). Finally, the individual subtypes are automatically derived by hierarchical clustering on these network representations. Results: We applied netMUG to a dataset containing genomic data and facial images to obtain BMI-informed multi-view strata and showed how it could be used for a refined obesity characterization. Benchmark analysis of netMUG on synthetic data with known strata of individuals indicated its superior performance compared with both baseline and benchmark methods for multi-view clustering. The clustering derived from netMUG achieved an adjusted Rand index of 1 with respect to the synthesized true labels. In addition, the real-data analysis revealed subgroups strongly linked to BMI and genetic and facial determinants of these subgroups. Discussion: netMUG provides a powerful strategy, exploiting individual-specific networks to identify meaningful and actionable strata. Moreover, the implementation is easy to generalize to accommodate heterogeneous data sources or highlight data structures.

PLEX.I: a tool to discover features in multiplex networks that reflect clinical variation

Molecular profiling technologies, such as RNA sequencing, offer new opportunities to better discover and understand the molecular networks involved in complex biological processes. Clinically important variations of diseases, or responses to treatment, are often reflected, or even caused, by the dysregulation of molecular interaction networks specific to particular network regions. In this work, we propose the R package PLEX.I, that allows quantifying and testing variation in the direct neighborhood of a given node between networks corresponding to different conditions or states. We illustrate PLEX.I in two applications in which we discover variation that is associated with different responses to tamoxifen treatment and to sex-specific responses to bacterial stimuli. In the first case, PLEX.I analysis identifies two known pathways i) that have already been implicated in the same context as the tamoxifen mechanism of action, and ii) that would have not have been identified using classical differential gene expression analysis.