EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

ViralMultiNet: A structure-aware multimodal framework for viral protein function prediction in wastewater surveillance

FuGuo Liu iD, TingLian Lai, WenXia Xu, GuoDong Li iD

PLOS One · 03.06.2026 · 21 (6) · e0349393

DOI: 10.1371/journal.pone.0349393 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS One, online erschienen am 03.06.2026. Verfasst von FuGuo Liu und 3 weiteren Autor:innen. Beteiligte Einrichtung: School of Statistics and Data Science, Xinjiang University of Finance and Economics, Urumqi, China. Hauptthemen laut MEDLINE: Wastewater, Viral Proteins, SARS-CoV-2. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 52 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Natural Science Foundation of Guangxi Province, Science and Technology Project of Guangxi.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate functional annotation of viral proteins is essential for genomic surveillance, yet rapid viral evolution causes “functional drift” that challenges conventional sequence-only models. These models often lack interpretability and struggle with fragmented sequences from complex environmental samples such as wastewater. We developed ViralMultiNet, a structure-aware multimodal framework that integrates multi-scale k-mer encodings (4–7-mers) with functional semantic embeddings derived from UniProt annotations. Using a curated Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) dataset of 66,011 samples from wastewater metagenomics (NCBI SRA: SRX28474964), we implemented gated multimodal fusion and triple knowledge distillation to transfer structural insights from a teacher to a student model. Model performance was evaluated via 5-fold cross-validation and external validation on emerging variants. Training efficiency was optimized using Low-Rank Adaptation and Flash Attention. ViralMultiNet achieved robust classification performance with a macro F1 score of 0.921 ± 0.004, accuracy of 0.928 ± 0.003, and AUC of 0.983 in cross-validation. The distilled student model matched teacher performance within a negligible margin (

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

WastewaterViral ProteinsSARS-CoV-2HumansCOVID-19Metagenomics

Substanzen

WastewaterViral Proteins

Autor:innen und Institutionen

  • School of Statistics and Data Science, Xinjiang University of Finance and Economics, Urumqi, China.
  • Department of Mathematics and Data Science, Changji University, Changji, China.
  • School of Mathematics and Computational Science, Guilin University of Electronic Technology, Guilin, China.

ORCID-Profile: FuGuo Liu, GuoDong Li

Förderung

  • Natural Science Foundation of Guangxi Province
  • Science and Technology Project of Guangxi

Zitieren

APA

Liu, F., Lai, T., Xu, W., & Li, G. (2026). ViralMultiNet: A structure-aware multimodal framework for viral protein function prediction in wastewater surveillance. PLOS One, 21(6), e0349393. https://doi.org/10.1371/journal.pone.0349393

Vancouver

Liu F, Lai T, Xu W, Li G. ViralMultiNet: A structure-aware multimodal framework for viral protein function prediction in wastewater surveillance. PLOS One. 2026;21(6):e0349393. doi:10.1371/journal.pone.0349393

BibTeX
@article{Liu2026,
  author = {Liu, FuGuo and Lai, TingLian and Xu, WenXia and Li, GuoDong},
  title = {ViralMultiNet: A structure-aware multimodal framework for viral protein function prediction in wastewater surveillance},
  journal = {PLOS One},
  year = {2026},
  volume = {21},
  number = {6},
  pages = {e0349393},
  doi = {10.1371/journal.pone.0349393}
}
Kostenlos starten →