{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,11]],"date-time":"2026-06-11T16:47:22Z","timestamp":1781196442482,"version":"3.54.1"},"reference-count":52,"publisher":"Public Library of Science (PLoS)","issue":"6","license":[{"start":{"date-parts":[[2026,6,3]],"date-time":"2026-06-03T00:00:00Z","timestamp":1780444800000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100004607","name":"Natural Science Foundation of Guangxi Province","doi-asserted-by":"publisher","award":["2026GXNSFAA00640099"],"award-info":[{"award-number":["2026GXNSFAA00640099"]}],"id":[{"id":"10.13039\/501100004607","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100004607","name":"Natural Science Foundation of Guangxi Province","doi-asserted-by":"publisher","award":["2026GXNSFAA00640099"],"award-info":[{"award-number":["2026GXNSFAA00640099"]}],"id":[{"id":"10.13039\/501100004607","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100004607","name":"Natural Science Foundation of Guangxi Province","doi-asserted-by":"publisher","award":["2026GXNSFAA00640099"],"award-info":[{"award-number":["2026GXNSFAA00640099"]}],"id":[{"id":"10.13039\/501100004607","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100004607","name":"Natural Science Foundation of Guangxi Province","doi-asserted-by":"publisher","award":["2026GXNSFAA00640099"],"award-info":[{"award-number":["2026GXNSFAA00640099"]}],"id":[{"id":"10.13039\/501100004607","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Science and Technology Project of Guangxi","award":["GuikeAD25069086"],"award-info":[{"award-number":["GuikeAD25069086"]}]},{"name":"Science and Technology Project of Guangxi","award":["GuikeAD25069086"],"award-info":[{"award-number":["GuikeAD25069086"]}]},{"name":"Science and Technology Project of Guangxi","award":["GuikeAD25069086"],"award-info":[{"award-number":["GuikeAD25069086"]}]},{"name":"Science and Technology Project of Guangxi","award":["GuikeAD25069086"],"award-info":[{"award-number":["GuikeAD25069086"]}]}],"content-domain":{"domain":["www.plosone.org"],"crossmark-restriction":false},"short-container-title":["PLoS One"],"abstract":"<jats:p>Accurate functional annotation of viral proteins is essential for genomic surveillance, yet rapid viral evolution causes \u201cfunctional drift\u201d that challenges conventional sequence-only models. These models often lack interpretability and struggle with fragmented sequences from complex environmental samples such as wastewater.<\/jats:p>\n                  <jats:p>We developed ViralMultiNet, a structure-aware multimodal framework that integrates multi-scale k-mer encodings (4\u20137-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.<\/jats:p>\n                  <jats:p>ViralMultiNet achieved robust classification performance with a macro F1 score of 0.921\u2009\u00b1\u20090.004, accuracy of 0.928\u2009\u00b1\u20090.003, and AUC of 0.983 in cross-validation. The distilled student model matched teacher performance within a negligible margin (&lt;0.003 F1 difference) while reducing training time by 40.4% (from 94.3 to 56.2 minutes per epoch). Interpretability analysis revealed that model attention peaks consistently aligned with experimentally validated functional domains of the SARS-CoV-2 Spike protein, including the receptor-binding domain (residues 319\u2013541), S1\/S2 cleavage site (681\u2013685), and fusion peptide (816\u2013835).<\/jats:p>\n                  <jats:p>ViralMultiNet offers a scalable, interpretable solution for viral protein function prediction. Its ability to generalize across variants and map attention to critical biological regions supports deployment in wastewater-based early warning systems, enhancing global pandemic preparedness.<\/jats:p>","DOI":"10.1371\/journal.pone.0349393","type":"journal-article","created":{"date-parts":[[2026,6,3]],"date-time":"2026-06-03T17:44:40Z","timestamp":1780508680000},"page":"e0349393","update-policy":"https:\/\/doi.org\/10.1371\/journal.pone.corrections_policy","source":"Crossref","is-referenced-by-count":1,"title":["ViralMultiNet: A structure-aware multimodal framework for viral protein function prediction in wastewater surveillance"],"prefix":"10.1371","volume":"21","author":[{"ORCID":"https:\/\/orcid.org\/0009-0002-3626-0106","authenticated-orcid":true,"given":"FuGuo","family":"Liu","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"TingLian","family":"Lai","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"WenXia","family":"Xu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1275-2982","authenticated-orcid":true,"given":"GuoDong","family":"Li","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,6,3]]},"reference":[{"key":"pone.0349393.ref001","doi-asserted-by":"crossref","first-page":"145124","DOI":"10.1016\/j.scitotenv.2021.145124","article-title":"Early warning of COVID-19 via wastewater-based epidemiology: potential and bottlenecks","volume":"767","author":"Y Zhu","year":"2021","journal-title":"Sci Total Environ"},{"key":"pone.0349393.ref002","doi-asserted-by":"crossref","first-page":"453","DOI":"10.1016\/j.trac.2018.06.004","article-title":"Wastewater-based epidemiology biomarkers: Past, present and future","volume":"105","author":"PM Choi","year":"2018","journal-title":"TrAC Trends in Analytical Chemistry"},{"issue":"6","key":"pone.0349393.ref003","doi-asserted-by":"crossref","first-page":"361","DOI":"10.1038\/s41579-023-00878-2","article-title":"The evolution of SARS-CoV-2","volume":"21","author":"PV Markov","year":"2023","journal-title":"Nat Rev Microbiol"},{"issue":"2","key":"pone.0349393.ref004","doi-asserted-by":"crossref","DOI":"10.1016\/S1473-3099(23)00813-7","article-title":"Virological characteristics of the SARS-CoV-2 JN.1 variant","volume":"24","author":"Y Kaku","year":"2024","journal-title":"Lancet Infect Dis"},{"issue":"9","key":"pone.0349393.ref005","doi-asserted-by":"crossref","first-page":"833","DOI":"10.1038\/nbt.3935","article-title":"Shotgun metagenomics, from sampling to analysis","volume":"35","author":"C Quince","year":"2017","journal-title":"Nat Biotechnol"},{"key":"pone.0349393.ref006","article-title":"HyenaDNA: Long-range genomic sequence modeling at single nucleotide resolution.","author":"E Nguyen","year":"2023"},{"key":"pone.0349393.ref007","doi-asserted-by":"crossref","unstructured":"Zvyagin M, Brace A, Hippe K, et al. GenSLMs: Genome-scale language models reveal SARS-CoV-2 evolutionary dynamics. bioRxiv [Preprint]. 2022.","DOI":"10.1101\/2022.10.10.511571"},{"issue":"4","key":"pone.0349393.ref008","doi-asserted-by":"crossref","first-page":"1234","DOI":"10.1093\/bioinformatics\/btz682","article-title":"BioBERT: a pre-trained biomedical language representation model for biomedical text mining","volume":"36","author":"J Lee","year":"2020","journal-title":"Bioinformatics"},{"key":"pone.0349393.ref009","article-title":"DNABERT-CLIP: A multi-modal foundation model for genomics and epigenomics.","author":"C Deng","year":"2023"},{"issue":"7","key":"pone.0349393.ref010","first-page":"1","article-title":"Survey on knowledge distillation for large language models","volume":"56","author":"R Yang","year":"2023","journal-title":"ACM Comput Surv"},{"issue":"1","key":"pone.0349393.ref011","doi-asserted-by":"crossref","first-page":"37","DOI":"10.1186\/s40168-020-00990-y","article-title":"VirSorter2: a multi-classifier, expert-guided approach to detect diverse DNA and RNA viruses","volume":"9","author":"J Guo","year":"2021","journal-title":"Microbiome"},{"key":"pone.0349393.ref012","doi-asserted-by":"crossref","DOI":"10.7717\/peerj-cs.2693","article-title":"Fake news detection: state-of-the-art review and advances with attention to Arabic language aspects","volume":"11","author":"ES Albtoush","year":"2025","journal-title":"PeerJ Comput Sci"},{"issue":"2","key":"pone.0349393.ref013","doi-asserted-by":"crossref","DOI":"10.1016\/j.cell.2020.02.058","article-title":"Structure, Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein","volume":"181","author":"AC Walls","year":"2020","journal-title":"Cell"},{"issue":"6498","key":"pone.0349393.ref014","doi-asserted-by":"crossref","first-page":"1499","DOI":"10.1126\/science.abc1560","article-title":"Structural basis for inhibition of the RNA-dependent RNA polymerase from SARS-CoV-2 by remdesivir","volume":"368","author":"W Yin","year":"2020","journal-title":"Science"},{"key":"pone.0349393.ref015","article-title":"Deep learning methods for protein structure prediction and design\u2014progress and emerging trends","volume":"5","author":"Y Qin","year":"2024","journal-title":"MedComm"},{"issue":"6","key":"pone.0349393.ref016","doi-asserted-by":"crossref","first-page":"1011","DOI":"10.1007\/s11633-024-1510-8","article-title":"A comprehensive survey on trustworthy graph neural networks: privacy, robustness, fairness, and explainability","volume":"21","author":"E Dai","year":"2024","journal-title":"Mach Intell Res"},{"issue":"10","key":"pone.0349393.ref017","article-title":"Interpretable multimodal fusion model for bridged histology and genomics","volume":"12","author":"Y Chen","year":"2025","journal-title":"Adv Sci"},{"issue":"5","key":"pone.0349393.ref018","doi-asserted-by":"crossref","first-page":"829","DOI":"10.1162\/neco_a_01273","article-title":"A Survey on Deep Learning for Multimodal Data Fusion","volume":"32","author":"J Gao","year":"2020","journal-title":"Neural Comput"},{"issue":"23","key":"pone.0349393.ref019","doi-asserted-by":"crossref","first-page":"4706","DOI":"10.3390\/electronics13234706","article-title":"An Adaptive Multimodal Fusion 3D Object Detection Algorithm for Unmanned Systems in Adverse Weather","volume":"13","author":"S Wang","year":"2024","journal-title":"Electronics"},{"issue":"12","key":"pone.0349393.ref020","first-page":"244","article-title":"Multimodal deep learning: a systematic review and taxonomy of architectures and fusion strategies","volume":"55","author":"E Valle","year":"2022","journal-title":"ACM Comput Surv"},{"issue":"2","key":"pone.0349393.ref021","first-page":"91","article-title":"Leveraging semantic text analysis to improve the performance of transformer based relation extraction","volume":"61","author":"MTC Evans","year":"2024","journal-title":"Inf Process Manag"},{"key":"pone.0349393.ref022","doi-asserted-by":"crossref","first-page":"4894","DOI":"10.1109\/TMM.2022.3183394","article-title":"Multimodal-Based and Aesthetic-Guided Narrative Video Summarization","volume":"25","author":"J Xie","year":"2023","journal-title":"IEEE Trans Multimedia"},{"issue":"1","key":"pone.0349393.ref023","first-page":"32","article-title":"Vision transformers for image classification: a comparative survey","volume":"13","author":"Y Zhang","year":"2023","journal-title":"Technologies"},{"issue":"6","key":"pone.0349393.ref024","first-page":"302","article-title":"Enhanced viral genome classification using large language models","volume":"16","author":"A Kandeil","year":"2023","journal-title":"Algorithms"},{"issue":"1","key":"pone.0349393.ref025","first-page":"1750","article-title":"VITAP: a high precision tool for DNA and RNA viral classification","volume":"16","author":"B Zhang","year":"2025","journal-title":"Nat Commun"},{"issue":"2","key":"pone.0349393.ref026","article-title":"Graph neural networks for single-cell omics data","volume":"24","author":"T Yu","year":"2023","journal-title":"Brief Bioinform"},{"key":"pone.0349393.ref027","first-page":"7345","article-title":"Multi-task multi-sensor fusion for 3D object detection.","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","author":"M Liang","year":"2019"},{"key":"pone.0349393.ref028","first-page":"2976","article-title":"Modular and Parameter-Efficient Multimodal Fusion with Prompting","author":"S Liang","year":"2022","journal-title":"Findings of ACL"},{"key":"pone.0349393.ref029","first-page":"120131","article-title":"MMAF Net: Multi view Multi stage Adaptive Fusion Network for 3D Object Detection in Adverse Weather","volume":"230","author":"W Zhang","year":"2023","journal-title":"Expert Syst Appl"},{"issue":"1","key":"pone.0349393.ref030","doi-asserted-by":"crossref","first-page":"164","DOI":"10.1186\/s12859-024-05773-y","article-title":"Orthogonal multimodality integration and clustering in single-cell data","volume":"25","author":"Y Liu","year":"2024","journal-title":"BMC Bioinformatics"},{"key":"pone.0349393.ref031","doi-asserted-by":"crossref","first-page":"17","DOI":"10.1007\/s42001-024-00345-9","article-title":"Open-source LLMs for text annotation: a practical guide","volume":"8","author":"M Alizadeh","year":"2024","journal-title":"J Comput Soc Sci"},{"key":"pone.0349393.ref032","doi-asserted-by":"crossref","first-page":"8279342","DOI":"10.1155\/2020\/8279342","article-title":"A Review of Multimodal Medical Image Fusion Techniques","volume":"2020","author":"B Huang","year":"2020","journal-title":"Comput Math Methods Med"},{"key":"pone.0349393.ref033","first-page":"112821","article-title":"A review: Deep learning for medical image segmentation using multi-modality fusion","volume":"138","author":"T Zhou","year":"2019","journal-title":"Expert Syst Appl"},{"issue":"2","key":"pone.0349393.ref034","doi-asserted-by":"crossref","first-page":"2627","DOI":"10.1007\/s40747-023-01289-9","article-title":"Bert-based graph unlinked embedding for sentiment analysis","volume":"10","author":"Y Jin","year":"2023","journal-title":"Complex Intell Syst"},{"issue":"6","key":"pone.0349393.ref035","doi-asserted-by":"crossref","first-page":"985","DOI":"10.1007\/s00779-024-01827-3","article-title":"Enhancing smart home interaction through multimodal command disambiguation","volume":"28","author":"T Cal\u00f2","year":"2024","journal-title":"Pers Ubiquit Comput"},{"issue":"1","key":"pone.0349393.ref036","first-page":"1","article-title":"A comprehensive survey on deep learning multi modal fusion: methods, technologies and applications","volume":"80","author":"T Jiao","year":"2020","journal-title":"Comput Mater Contin"},{"issue":"3","key":"pone.0349393.ref037","first-page":"1","article-title":"Deep multimodal data fusion for robotic applications: a comprehensive survey","volume":"56","author":"Z Qian","year":"2024","journal-title":"ACM Comput Surv"},{"key":"pone.0349393.ref038","doi-asserted-by":"crossref","first-page":"104731","DOI":"10.1016\/j.jbi.2024.104731","article-title":"Augmenting biomedical named entity recognition with general-domain resources","volume":"159","author":"Y Yin","year":"2024","journal-title":"J Biomed Inform"},{"key":"pone.0349393.ref039","doi-asserted-by":"crossref","first-page":"126866","DOI":"10.1016\/j.neucom.2023.126866","article-title":"A review of multimodal emotion recognition from datasets, preprocessing, features, and fusion methods","volume":"561","author":"Y Zhang","year":"2023","journal-title":"Neurocomputing"},{"key":"pone.0349393.ref040","doi-asserted-by":"crossref","first-page":"102900","DOI":"10.1016\/j.artmed.2024.102900","article-title":"Transformers and large language models in healthcare: A review","volume":"154","author":"S Nerella","year":"2024","journal-title":"Artif Intell Med"},{"key":"pone.0349393.ref041","doi-asserted-by":"crossref","first-page":"105253","DOI":"10.1016\/j.compbiomed.2022.105253","article-title":"A review on multimodal medical image fusion: Compendious analysis of medical modalities, multimodal databases, fusion techniques and quality metrics","volume":"144","author":"MA Azam","year":"2022","journal-title":"Comput Biol Med"},{"key":"pone.0349393.ref042","first-page":"1","article-title":"Survey on sentiment analysis: evolution of research methods and topics","author":"J Cui","year":"2023","journal-title":"Artif Intell Rev"},{"issue":"4","key":"pone.0349393.ref043","doi-asserted-by":"crossref","DOI":"10.1002\/widm.1458","article-title":"A review on data fusion in multimodal learning analytics and educational data mining","volume":"12","author":"W Chango","year":"2022","journal-title":"WIREs Data Min & Knowl"},{"key":"pone.0349393.ref044","first-page":"1","article-title":"A systematic review of social network sentiment analysis with comparative study of ensemble-based techniques","author":"D Tiwari","year":"2023","journal-title":"Artif Intell Rev"},{"issue":"7","key":"pone.0349393.ref045","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1145\/3648471","article-title":"Utilizing BERT for Information Retrieval: Survey, Applications, Resources, and Challenges","volume":"56","author":"J Wang","year":"2024","journal-title":"ACM Comput Surv"},{"issue":"3","key":"pone.0349393.ref046","first-page":"1","article-title":"Deep multimodal data fusion: a comprehensive survey","volume":"56","author":"Z Qian","year":"2024","journal-title":"ACM Comput Surv"},{"issue":"1","key":"pone.0349393.ref047","doi-asserted-by":"crossref","first-page":"35","DOI":"10.54254\/2755-2721\/101\/20240946","article-title":"Transformer models in text summarization","volume":"101","author":"X Fu","year":"2024","journal-title":"Appl Comput Eng"},{"issue":"5","key":"pone.0349393.ref048","doi-asserted-by":"crossref","first-page":"551","DOI":"10.1016\/j.ijhcs.2012.12.004","article-title":"An exploration of pen tail gestures for interactions","volume":"71","author":"F Tian","year":"2013","journal-title":"Int J Hum Comput Stud"},{"issue":"4","key":"pone.0349393.ref049","doi-asserted-by":"crossref","first-page":"863","DOI":"10.1007\/s10506-023-09374-7","article-title":"Bringing order into the realm of Transformer-based language models for artificial intelligence and law","volume":"32","author":"CM Greco","year":"2023","journal-title":"Artif Intell Law"},{"issue":"8","key":"pone.0349393.ref050","doi-asserted-by":"crossref","first-page":"1514","DOI":"10.3390\/land12081514","article-title":"A Systematic Review of Disaster Management Systems: Approaches, Challenges, and Future Directions","volume":"12","author":"SM Khan","year":"2023","journal-title":"Land"},{"key":"pone.0349393.ref051","doi-asserted-by":"crossref","first-page":"102755","DOI":"10.1016\/j.inffus.2024.102755","article-title":"Natural language processing in finance: A survey","volume":"115","author":"K Du","year":"2025","journal-title":"Information Fusion"},{"key":"pone.0349393.ref052","first-page":"102456","article-title":"Multimodal interaction in virtual reality supported education: a systematic review","volume":"171","author":"H Hu","year":"2024","journal-title":"Int J Hum Comput Stud"}],"container-title":["PLOS One"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0349393","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,3]],"date-time":"2026-06-03T17:44:49Z","timestamp":1780508689000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0349393"}},"subtitle":[],"editor":[{"given":"Ot\u00e1vio Augusto","family":"Chaves","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,6,3]]},"references-count":52,"journal-issue":{"issue":"6","published-online":{"date-parts":[[2026,6,3]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pone.0349393","relation":{},"ISSN":["1932-6203"],"issn-type":[{"value":"1932-6203","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,6,3]]}}}