Predictability of influenza virus evolution

Richard Neher
Biozentrum & SIB, University of Basel

slides at

Human seasonal influenza viruses

slide by Trevor Bedford

Positive tests for influenza in the USA by week

Data by the US CDC

Genomic analysis to reconstruct pathogen spread and evolution

Link genotypic and phenotypic changes: immune escape, drug resistance, host adaptation

Influenza A/H3N2

  • Influenza viruses evolve to avoid human immunity
  • Vaccines need frequent updates

Vaccine strain selection schedule

Klingen and McHardy, Trends in Microbiology

Beyond tracking: can we predict?

Can we pick a "winner"?

Traditional approach: evolutionary race of viruses

  • Speed of adaptation is logarithmic in population size
  • Single mutations are easy to find, many such mutation are needed for success
  • Different models have universal emerging properties
RN, Annual Reviews, 2013; Desai & Fisher; Brunet & Derride; Kessler & Levine

Neutral/Kingman coalescent

strong selection

Bolthausen-Sznitman Coalescent

RN, Hallatschek, PNAS, 2013; see also Brunet and Derrida, PRE, 2007; Desai, Walczak, Fisher, Genetics, 2013

Burst in the tree ↔ high fitness

Predicting evolution

Given the branching pattern:

  • can we predict fitness?
  • pick the closest relative of the future?
RN, Russell, Shraiman, eLife, 2014

Fitness inference from trees

$$P(\mathbf{x}|T) = \frac{1}{Z(T)} p_0(x_0) \prod_{i=0}^{n_{int}} g(x_{i_1}, t_{i_1}| x_i, t_i)g(x_{i_2}, t_{i_2}| x_i, t_i)$$
RN, Russell, Shraiman, eLife, 2014

Predicting an optimal representative

RN, Russell, Shraiman, eLife, 2014

Moderate prediction success

  • But is a random strain for previous years a sensible null?
  • Does this work for the right reason?
RN, Russell, Shraiman, eLife, 2014

Predicting the distribution in sequence space

Huddleston et al, eLife, 2020

Follow up investigations show very modest predictability

  • Optimal transport distances between prediction and observation -- less than one amino acid gained
Huddleston et al, eLife, 2020

Do A/H3N2 mutations have inertia?

Barrat-Charlaix et al, 2020

Complicated "Ecology" of host and pathogen

  • Current approaches focus on the virus population
  • Models are predicated on identifying a 'winner'
  • Instead, dynamics might be slaved to host immunity:
    → exposure history and waning determine population immunity, viruses fill whatever niche there is
  • For prediction, host data would be at least as important as viral dynamics
  • Prediction horizon is limited to within season dynamics that equilibrates the host-pathogen immunity landscape
  • Viral adaptations are niche specific and loose their benefit after equilibration.
Barrat-Charlaix et al, 2020

Diversity in immune responses

Lee et al, 2019

Influenza and Theory acknowledgments

  • Boris Shraiman
  • Colin Russell
  • Trevor Bedford
  • Pierre Barrat
  • John Huddleston

  • All the NICs and WHO CCs that provide influenza sequence data
  • The WHO CCs in London and Atlanta for providing titer data


Trevor Bedford and his lab -- terrific collaboration since 2014

especially James Hadfield, Emma Hodcroft, Ivan Aksamentov, Cornelius Roemer, Moira Zuber, and John Huddleston

Data we analyze are contributed by scientists from all over the world

Data are shared and curated by GISAID

Fixation probability is quasi-neutral

A/H3N2 influenza
Simulations with increasing interference
Barrat-Charlaix et al, 2020