Scholarlink

Background <p>Human-built environment microbiomes mediate pathogen persistence and antimicrobial resistance (AMR) circulation, yet their ecological organization and resilience remain poorly quantified. Hospitals, sewage systems, ambulances, and public transport form interconnected microbial networks where contamination potential and compositional stability define biosurveillance risk. Understanding these dynamics requires genome-resolved frameworks capable of linking community composition to ecological behavior.</p> Methods <p>We analyzed 767 publicly available Illumina metagenomes from four urban environments using the GRUMB workflow. Quality-filtered reads were assembled into 10,834 metagenome-assembled genomes (MAGs) and dereplicated into 1542 species-level representatives. Functional annotation with CARD and VFDB identified ARG- and VF-carrying species, producing a genome-resolved abundance matrix used for ecological and predictive modeling. Alpha and beta diversity, indicator taxa, and prevalence were assessed in R, while machine learning (Random Forest, scikit-learn) achieved a nested cross-validation balanced accuracy of 0.97 ± 0.01. Synthetic donor–recipient simulations (<i>α</i> = 0–1) implemented in Python modeled compositional blending, entropy-based uncertainty, and Minimal Detectable Contamination (MDC) thresholds.</p> Results <p>Microbial communities exhibited strong environment-specific structure (PERMANOVA <i>R</i><sup>2</sup> = 0.12, <i>p</i> &lt; 0.001). Hospital sewage contained the highest richness and compositional heterogeneity, whereas ambulances and hospital environments showed low-diversity, surface-filtered microbiomes. Machine learning identified consistent ecological predictors (<i>Pseudomonas_E fragi</i>, <i>Sphingomonas sp000797515</i>, <i>Acinetobacter variabilis</i>, <i>Roseomonas mucosa</i>) that delineated environmental identity. Synthetic blending revealed a directional source–sink hierarchy with hospital sewage acting as the primary donor (MDC = 0.2–0.3), while hospital environments displayed the greatest compositional resilience (MDC ≥ 0.8). Entropy-based uncertainty analysis identified tipping zones (<i>α</i> = 0.3–0.5), and dominance mapping highlighted hospital environments as stabilizing ecological nodes. WHO-priority pathogens (<i>Acinetobacter baumannii</i>, <i>Klebsiella pneumoniae</i>, <i>Escherichia coli</i>) occupied central positions in the network, bridging environmental and clinical compartments.</p> Conclusions <p>This genome-resolved and simulation-driven framework reveals a directional microbial continuum across urban infrastructures governed by dominance, resilience, and clinical connectivity. Hospital sewage functions as a microbial donor, while hospital environments act as ecological stabilizers anchoring built-environment microbiomes. These findings advance biosurveillance from descriptive profiling to predictive ecological modeling, offering quantitative metrics for risk-informed infrastructure design.</p> <p><MediaObject ID="MOESM6"> <VideoObject FileRef="MediaObjects/40168_2025_2315_MOESM6_ESM.mp4" VideoID="-kZevqo-edgZ8sPpqBPB72"> <Caption Language="En" xml:lang="en"> <CaptionContent> <p>Video Abstract</p> </CaptionContent> </Caption> </VideoObject> </MediaObject></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Genome-resolved surveillance and predictive ecological risk modeling of urban microbiomes

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月23日

查看全文
  • Suleiman Aminu,
  • AbdulAziz Ascandari,
  • Morad M. Mokhtar,
  • Achraf El Allali,
  • Rachid Benhida,
  • Rachid Daoud

摘要

Background

Human-built environment microbiomes mediate pathogen persistence and antimicrobial resistance (AMR) circulation, yet their ecological organization and resilience remain poorly quantified. Hospitals, sewage systems, ambulances, and public transport form interconnected microbial networks where contamination potential and compositional stability define biosurveillance risk. Understanding these dynamics requires genome-resolved frameworks capable of linking community composition to ecological behavior.

Methods

We analyzed 767 publicly available Illumina metagenomes from four urban environments using the GRUMB workflow. Quality-filtered reads were assembled into 10,834 metagenome-assembled genomes (MAGs) and dereplicated into 1542 species-level representatives. Functional annotation with CARD and VFDB identified ARG- and VF-carrying species, producing a genome-resolved abundance matrix used for ecological and predictive modeling. Alpha and beta diversity, indicator taxa, and prevalence were assessed in R, while machine learning (Random Forest, scikit-learn) achieved a nested cross-validation balanced accuracy of 0.97 ± 0.01. Synthetic donor–recipient simulations (α = 0–1) implemented in Python modeled compositional blending, entropy-based uncertainty, and Minimal Detectable Contamination (MDC) thresholds.

Results

Microbial communities exhibited strong environment-specific structure (PERMANOVA R2 = 0.12, p < 0.001). Hospital sewage contained the highest richness and compositional heterogeneity, whereas ambulances and hospital environments showed low-diversity, surface-filtered microbiomes. Machine learning identified consistent ecological predictors (Pseudomonas_E fragi, Sphingomonas sp000797515, Acinetobacter variabilis, Roseomonas mucosa) that delineated environmental identity. Synthetic blending revealed a directional source–sink hierarchy with hospital sewage acting as the primary donor (MDC = 0.2–0.3), while hospital environments displayed the greatest compositional resilience (MDC ≥ 0.8). Entropy-based uncertainty analysis identified tipping zones (α = 0.3–0.5), and dominance mapping highlighted hospital environments as stabilizing ecological nodes. WHO-priority pathogens (Acinetobacter baumannii, Klebsiella pneumoniae, Escherichia coli) occupied central positions in the network, bridging environmental and clinical compartments.

Conclusions

This genome-resolved and simulation-driven framework reveals a directional microbial continuum across urban infrastructures governed by dominance, resilience, and clinical connectivity. Hospital sewage functions as a microbial donor, while hospital environments act as ecological stabilizers anchoring built-environment microbiomes. These findings advance biosurveillance from descriptive profiling to predictive ecological modeling, offering quantitative metrics for risk-informed infrastructure design.

Video Abstract