Using homologous network to identify reassortment risk in H5Nx avian influenza viruses

Ruihao Gong, Zijian Feng, Yanyun ZhangPLOS Computational Biology 2025 · DOI

TL;DR

This paper introduces a novel network-based approach to quantify and rank reassortment risk in H5Nx avian influenza viruses. By constructing a "homologous network" of viral genomes, the authors identify key host populations (domestic poultry in China and wild birds in North America/Europe) that are critical for future surveillance and intervention efforts against these dangerous viruses.

What Problem Does This Paper Address?

Influenza A viruses (IAVs), particularly H5Nx subtypes, pose a significant threat to global health due to their ability to rapidly evolve through reassortment. This process, where viral segments are exchanged between different strains, allows viruses to adapt to new hosts, evade immunity, and increase pathogenicity. However, the complexity of IAV diversity, host range, and the sporadic nature of reassortment make it challenging to accurately assess and predict reassortment risk. Existing methods often struggle with large datasets and the intricate interplay of factors involved, hindering effective preparedness and surveillance strategies.

Methods

The core of this study is the development and application of an "IAVs homologous network" approach. Here's a breakdown:

  • Network Construction: The authors collected a massive dataset of IAV whole-genome sequences. They then inferred phylogenetic trees for each of the eight gene segments to define viral lineages. A "homologous network" was constructed where each virus strain is a node, and a link exists between two viruses if they share genomic homogeneity in at least one segment. The strength of this link (weight) is determined by the number of shared segments.
  • Community Detection for Reassortment Risk: To quantify reassortment risk, the researchers applied a Hierarchical Link Clustering (HLC) algorithm to identify "communities" within the network. These communities represent groups of viruses with similar genomic compositions. The number of communities a virus belongs to serves as an indicator of its reassortment risk, reflecting its genomic diversity and history of segment exchange.
  • Host-Specific Risk Analysis: The authors then analyzed the reassortment risk associated with different host types (e.g., domestic poultry, wild birds) across various regions. They also explored how removing specific host types from the network would impact community structure, inferring their importance in facilitating reassortment and cross-species transmission.

Datasets

  • GISAID Database: The study utilized a comprehensive dataset of influenza A virus whole-genome sequences downloaded from the Global Initiative on Sharing All Influenza Data (GISAID) database, covering data up to 2023.
  • Downsampled IAVs Dataset: To mitigate biases from uneven surveillance, the raw dataset was downsampled to create a more equitable distribution of sequences across regions, hosts, and subtypes. The final H5Nx dataset comprised 3420 sequences.
  • Simulated Datasets: For methodological validation, the authors generated simulated IAVs datasets with varying reassortment rates and also created randomized datasets by shuffling existing data to test for false positives.

Key Insights and Ideas

  • Network as a "Reassortment History": The homologous network isn't just a static representation; it implicitly encodes the reassortment history of viruses by linking strains with shared genomic components.
  • Community Count as a Risk Metric: The number of communities a virus belongs to is a clever proxy for its reassortment risk, capturing its genomic complexity and potential for further exchange.
  • Identifying "Super-Spreader" Hosts: The network approach effectively highlights specific host populations that are disproportionately involved in driving reassortment, providing actionable targets for surveillance.
  • Regional Differences in Risk: Reassortment risk isn't uniform; it varies significantly by region, with distinct host types playing dominant roles in China versus North America/Europe.

Key Results

  • Diverse Reassortment Risks: The analysis revealed that different H5Nx viruses exhibit varying reassortment risks, as indicated by the number of communities they belong to.
  • High-Risk Hosts Identified:
    • In China, domestic poultry (specifically Galliformes, like chickens) were identified as having the highest reassortment risk.
  • In North America and Europe, wild Anseriformes (like ducks and geese) were the primary contributors to reassortment risk.
  • Surveillance Targets: The study predicts that these high-risk host populations (domestic poultry in China, wild birds in North America and Europe) are crucial targets for future H5Nx surveillance efforts.
  • Methodological Robustness: The network approach was validated using simulated datasets, demonstrating its reliability in identifying reassortment risk without introducing false positives.

Limitations and Future Work

  • Surveillance Bias: The authors acknowledge that despite downsampling, the inherent limitations and biases in global IAV surveillance (e.g., only a fraction of outbreaks are sequenced) can still influence the network construction and risk assessment.
  • Host Grouping Simplification: Categorizing wild avian hosts into broad groups might oversimplify the complex migration behaviors and interactions that influence reassortment.
  • Regional Granularity: The broad regional classifications (China, North America, Europe) might mask finer-scale geographic variations in reassortment dynamics.
  • Human Role: The role of humans in contributing to H5Nx reassortment is noted as potentially overlooked due to active medical interventions (sampling).

Future work could involve refining host classifications, incorporating more detailed epidemiological data, and further investigating the specific mechanisms by which identified host populations facilitate reassortment. The authors emphasize that their framework provides a valuable tool for designing more targeted surveillance and preparedness strategies.

Summary updated

BibTeX
@article{gong2025using,
  author = {Ruihao Gong and Zijian Feng and Yanyun Zhang},
  title = {{Using homologous network to identify reassortment risk in H5Nx avian influenza viruses}},
  journal = {PLOS Computational Biology},
  year = {2025},
  doi = {10.1371/journal.pcbi.1013301},
}
Figure