Language models learn to represent antigenic properties of human influenza A(H3) virus
Abstract Given that influenza vaccine effectiveness depends on a good antigenic match between the vaccine and circulating viruses, it is important to assess the antigenic properties of newly emerging variants continuously. With the increasing application of real-time pathogen genomic surveillance, a...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-07-01
|
| Series: | Scientific Reports |
| Online Access: | https://doi.org/10.1038/s41598-025-03275-2 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849238652220080128 |
|---|---|
| author | Francesco Durazzi Marion P. G. Koopmans Ron A. M. Fouchier Daniel Remondini |
| author_facet | Francesco Durazzi Marion P. G. Koopmans Ron A. M. Fouchier Daniel Remondini |
| author_sort | Francesco Durazzi |
| collection | DOAJ |
| description | Abstract Given that influenza vaccine effectiveness depends on a good antigenic match between the vaccine and circulating viruses, it is important to assess the antigenic properties of newly emerging variants continuously. With the increasing application of real-time pathogen genomic surveillance, a key question is if antigenic properties can reliably be predicted from influenza virus genomic information. Based on validated linked datasets of influenza virus genomic and wet lab experimental results, in silico models may be of use to learn to predict immune escape of variants of interest starting from the protein sequence only. In this study, we compared several machine-learning methods to reconstruct antigenic map coordinates for HA1 protein sequences of influenza A(H3N2) virus, to rank substitutions responsible for major antigenic changes, and to recognize variants with novel antigenic properties that may warrant future vaccine updates. Methods based on deep learning language models (BiLSTM and ProtBERT) and more classical approaches based solely on genetic distances and physicochemical properties of amino acid sequences had comparable performances over the coarser features of the map, but the first two performed better over fine-grained features like single amino acid-driven antigenic change and in silico deep mutational scanning experiments to rank the substitutions with the largest impact on antigenic properties. Given that the best performing model that produces protein embeddings is agnostic to the specific pathogen, the presented approach may be applicable to other pathogens. |
| format | Article |
| id | doaj-art-d6428ff5f1e54544b319d9ec54fd5d23 |
| institution | Kabale University |
| issn | 2045-2322 |
| language | English |
| publishDate | 2025-07-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj-art-d6428ff5f1e54544b319d9ec54fd5d232025-08-20T04:01:26ZengNature PortfolioScientific Reports2045-23222025-07-0115111010.1038/s41598-025-03275-2Language models learn to represent antigenic properties of human influenza A(H3) virusFrancesco Durazzi0Marion P. G. Koopmans1Ron A. M. Fouchier2Daniel Remondini3Department of Physics and Astronomy, University of BolognaDepartment of Viroscience, Erasmus Medical CentreDepartment of Viroscience, Erasmus Medical CentreDepartment of Physics and Astronomy, University of BolognaAbstract Given that influenza vaccine effectiveness depends on a good antigenic match between the vaccine and circulating viruses, it is important to assess the antigenic properties of newly emerging variants continuously. With the increasing application of real-time pathogen genomic surveillance, a key question is if antigenic properties can reliably be predicted from influenza virus genomic information. Based on validated linked datasets of influenza virus genomic and wet lab experimental results, in silico models may be of use to learn to predict immune escape of variants of interest starting from the protein sequence only. In this study, we compared several machine-learning methods to reconstruct antigenic map coordinates for HA1 protein sequences of influenza A(H3N2) virus, to rank substitutions responsible for major antigenic changes, and to recognize variants with novel antigenic properties that may warrant future vaccine updates. Methods based on deep learning language models (BiLSTM and ProtBERT) and more classical approaches based solely on genetic distances and physicochemical properties of amino acid sequences had comparable performances over the coarser features of the map, but the first two performed better over fine-grained features like single amino acid-driven antigenic change and in silico deep mutational scanning experiments to rank the substitutions with the largest impact on antigenic properties. Given that the best performing model that produces protein embeddings is agnostic to the specific pathogen, the presented approach may be applicable to other pathogens.https://doi.org/10.1038/s41598-025-03275-2 |
| spellingShingle | Francesco Durazzi Marion P. G. Koopmans Ron A. M. Fouchier Daniel Remondini Language models learn to represent antigenic properties of human influenza A(H3) virus Scientific Reports |
| title | Language models learn to represent antigenic properties of human influenza A(H3) virus |
| title_full | Language models learn to represent antigenic properties of human influenza A(H3) virus |
| title_fullStr | Language models learn to represent antigenic properties of human influenza A(H3) virus |
| title_full_unstemmed | Language models learn to represent antigenic properties of human influenza A(H3) virus |
| title_short | Language models learn to represent antigenic properties of human influenza A(H3) virus |
| title_sort | language models learn to represent antigenic properties of human influenza a h3 virus |
| url | https://doi.org/10.1038/s41598-025-03275-2 |
| work_keys_str_mv | AT francescodurazzi languagemodelslearntorepresentantigenicpropertiesofhumaninfluenzaah3virus AT marionpgkoopmans languagemodelslearntorepresentantigenicpropertiesofhumaninfluenzaah3virus AT ronamfouchier languagemodelslearntorepresentantigenicpropertiesofhumaninfluenzaah3virus AT danielremondini languagemodelslearntorepresentantigenicpropertiesofhumaninfluenzaah3virus |