Visualization of seasonal influenza antigenic evolution

Seasonal influenza kills 100,000s
of people every year

Photo: WHO / Tom Pietrasik

Surface proteins determine
influenza virus subtypes like H3N2

Hemagglutinin enables infection

Hemagglutinin is the primary target of our immune system

Mutations in hemagglutinin allow viruses to escape existing immunity

Mutations in hemagglutinin allow viruses to escape existing immunity

The Global Influenza Surveillance and Response System tracks influenza year-round

Surveillance groups sequence
the genetic code of viruses

Visualizing genomes as an inferred genealogy reveals related groups of viruses

Which single virus should be in the next vaccine?

Experimental assays identify viruses
that could escape our immunity

Assays provide antigenic distances

Historically, distances were viewed as tables

Visualization as tables does not scale well

Antigenic cartography reduces
high dimensionality

Bayesian antigenic cartography integrates
assays and genetic data

Antigenic trees map assays to phylogenies
and infer missing data

Antigenic trees synthesize all assay data
into a single visualization

Scatterplot trees encode antigenic distance
on a positional axis

Matrix view emphasizes coverage by
each serum of test viruses by clade

Distribution view communicates uncertainty about coverage of test viruses by clade

Distributions with raw data reveal details about coverage of test viruses by clade

Interactive visualizations support
multiple views on the data

Composite tree and measurements views provide all information required for decision making

Implementation complexity hinders
domain-specific visualizations

Thank you!

Continue the conversation