Graph-guided multi-task sparse learning model: a method for identifying antigenic variants of influenza A(H3N2) virus

Area of Science:

• Virology
• Immunology
• Computational Biology

Background:

• Influenza virus antigenic variants emerge continuously, necessitating regular vaccine updates.
• Current methods for identifying antigenic variants are slow, expensive, and generate noisy data.
• Existing serologic methods face challenges in data interpretation and integration due to protocol variations.

Purpose of the Study:

• To develop a novel, efficient method for identifying influenza antigenic variants.
• To overcome the limitations of traditional serologic assays for vaccine strain selection.
• To enable real-time, large-scale characterization of influenza antigenic profiles.

Main Methods:

• Developed a Graph-Guided Multi-Task Sparse Learning (GG-MTSL) model.
• Utilized multi-sourced serologic data to learn antigenicity-associated mutations.
• Applied the GG-MTSL model to influenza H3N2 hemagglutinin sequences.

Main Results:

• The GG-MTSL model enables rapid characterization of antigenic profiles.
• Successfully identified influenza antigenic variants in real time and on a large scale.
• Demonstrated that sequences from clinical samples can minimize culture-adaptation biases.

Conclusions:

• GG-MTSL offers a robust solution for influenza antigenic variant identification.
• The method facilitates timely selection of vaccine strains.
• GG-MTSL provides a scalable and efficient approach for influenza surveillance.