Temporal Analysis of the Relative Transmission Fitness of Seasonal Influenza Subtypes Using the Differential Population Growth Rate Model
Abstract/Description/Artist Statement
The rapid evolution and continuous emergence of new Influenza A lineages pose a significant challenge to global public health and vaccine strain selection. Accurately predicting the evolutionary trajectory of circulating clades requires robust, mathematically sound measurements of viral fitness. However, applying relative fitness metrics to global surveillance data is often difficult because different flu variants often circulate in different regions at different times, creating gaps in the data.
To overcome this, we utilized a sliding-window data-driven pairwise comparison method; Differential Population Growth Rate (DPGR) model, which minimizes sampling bias by using co-circulating viral strains as internal controls, to calculate the pairwise relative fitness of Influenza A variants based on frequency data. We focused on variants that have been dominating across multiple regions and flu seasons.
By estimating the relative transmission fitness of competing flu variants, this pipeline successfully identified the dominant lineages across multiple seasons. Additionally, we were able to link these quantitative fitness advantages to real-world outcomes, showing how the transmission fitness of the various variants correspond with the strains selected for seasonal vaccines. This confirms the pipeline's ability to accurately reflect real-world viral dynamics and seasonal dominance.
This DPGR pipeline generates a quantitative metric of viral pairwise transmission patterns. For future work, the absolute fitness scores will be integrated with aligned genomic sequences to train neural network models. Ultimately, this deep learning model aims to map raw genetic changes directly to epidemiological fitness, providing a predictive tool for anticipating the dominance of future influenza variants.
Keywords: DPGR, relative fitness, influenza, neural network, vaccine selection, predictive modelling, public health, genomic surveillance
Faculty Advisor/Mentor
Hong Qin
Faculty Advisor/Mentor Email
hqin@odu.edu
Faculty Advisor/Mentor Department
Department of Computer Science
College/School Affiliation
College of Sciences
Student Level Group
Graduate/Professional
Presentation Type
Poster
Temporal Analysis of the Relative Transmission Fitness of Seasonal Influenza Subtypes Using the Differential Population Growth Rate Model
The rapid evolution and continuous emergence of new Influenza A lineages pose a significant challenge to global public health and vaccine strain selection. Accurately predicting the evolutionary trajectory of circulating clades requires robust, mathematically sound measurements of viral fitness. However, applying relative fitness metrics to global surveillance data is often difficult because different flu variants often circulate in different regions at different times, creating gaps in the data.
To overcome this, we utilized a sliding-window data-driven pairwise comparison method; Differential Population Growth Rate (DPGR) model, which minimizes sampling bias by using co-circulating viral strains as internal controls, to calculate the pairwise relative fitness of Influenza A variants based on frequency data. We focused on variants that have been dominating across multiple regions and flu seasons.
By estimating the relative transmission fitness of competing flu variants, this pipeline successfully identified the dominant lineages across multiple seasons. Additionally, we were able to link these quantitative fitness advantages to real-world outcomes, showing how the transmission fitness of the various variants correspond with the strains selected for seasonal vaccines. This confirms the pipeline's ability to accurately reflect real-world viral dynamics and seasonal dominance.
This DPGR pipeline generates a quantitative metric of viral pairwise transmission patterns. For future work, the absolute fitness scores will be integrated with aligned genomic sequences to train neural network models. Ultimately, this deep learning model aims to map raw genetic changes directly to epidemiological fitness, providing a predictive tool for anticipating the dominance of future influenza variants.
Keywords: DPGR, relative fitness, influenza, neural network, vaccine selection, predictive modelling, public health, genomic surveillance