Sequence-based detection of emerging antigenically novel influenza A viruses

Area of Science:

• Virology
• Computational Biology
• Immunology

Background:

• Detecting evolutionary changes in influenza A (H3N2) virus antigenicity is crucial for effective vaccine design.
• Current methods for identifying antigenic transitions can be labor-intensive and costly.

Purpose of the Study:

• To introduce Novel Influenza Virus A Detector (NIAViD), an unsupervised machine learning tool for identifying antigenic transitions in influenza A (H3N2) viruses.
• To evaluate NIAViD's performance in detecting these transitions using HA1 sequences and physico-chemical properties.

Main Methods:

• Utilized an unsupervised machine learning approach with HA1 sequences and physico-chemical properties.
• Assessed NIAViD's sensitivity in training (88.9%) and validation (72.7%) datasets.
• Investigated the role of the Boman's index in enhancing the detection of antigenic shifts.

Main Results:

• NIAViD demonstrated superior sensitivity compared to the uncalibrated null model (33.3%).
• The tool successfully identified novel antigenic clusters and pinpointed sites of significant antigenic change.
• NIAViD does not require traditional laboratory assays, reducing bias and cost.

Conclusions:

• NIAViD offers a promising, efficient method for detecting influenza virus antigenic transitions.
• This tool can augment current surveillance networks, providing timely insights for updated vaccine development.
• NIAViD supports informed decision-making for influenza vaccine strategies.