Gene transfer & pathogenicity: a big data approach to

influenza macroevolution

Paul AgapowDept. of Genomics of Common Disease

p.agapow@ic.ac.uk

October 2015

The dull yet necessary background facts

• A respiratory pathogen

• Negative-sense ssRNA virus

• 3 types (A, B, C)

• 8 chromosomes encoding 11 proteins

Cliches of influenza evolution

Shift

• Acquisition of large functional units via horizontal gene transfer

• Sudden adaptive leaps

• Infrequent (?) but fast

• Pandemics

The success of a strain of influenza depends on its ability to subvert or subdue host immunity, which in turn depends on evolution:

Drift

• Conventional molecular mutation

• Stepwise / incremental

• Continual but slow

• Seasonal flu

Shift happens

The problem

Reassortment

• Only a potential or freak event?

• If not, under what circumstances?

Recombination

• “...plays an important role in influenza evolution...”

• “...never happens...”

• Does it happen?

• Where & when does it happen?

• What happens?

Detecting HGT

Ordinarily the genes in two related organisms have the same most recent common ancestor. HGT disrupts this.

Correlating evolutionary distances

Construct "sequences" composed just of 3rd positions from coding regions. Then:

● For every pair of segments (HA vs NA, HA vs PA, etc):

● For every pair of isolates:

1. extract distance between homologous segments

2. graph the distance for gene A versus gene B

Where gene A and gene B share the same time to their MRCA, the observed evolution in A should be proportional to that in B.

Visualising correlation of evolutionary distances

The distance between two isolates is ΔA = kΔ

B+ x

The datasets

For now

• 67 swine-lineage H1N1

• 167 avian-lineage H5N1

• 525 human-lineage H3N2

Use for

• Testing reassortment between segments

• Testing recombination within segments by splitting into two

Does it work (in humans)?

Does it work (in pigs)?

Does it work (in birds)?

Recombination?

No (probably).

How to quantify reassortment?

With great difficulty ...

• Limited to comparisons

• What is a reassortment anyway?

• What is a rate?

Ad-hoc method

• Count populations (3+) of similar likelihood

• Measure “years” in phylogeny

• Measure relative population size as “success”

Reassortment across hosts and strains

Reassortment across segments

Conclusions

• There is no significant recombination

• Reassortment occurs in all compartments but varies

• Humans < pigs < birds

• structural < antigenic & replication

• More persistent in birds?

Open questions & next steps

• More (segments / strains / hosts)

• Wherefore recombination?

• What is driving the different dynamics?

• Unknown implication of sampling

• How to best compare host & strains?

• Integration into modelling