Phylogenetics III: Zika virus outbreak investigation

Phylogeography

This practical is a continuation of the previous page.

With access to good metadata such as geographic sampling locations and host species, one can use the phylogeny to investigate the origin and spread of the outbreak geographically and between host species.

The Google Docs spreadsheet presented earlier contains metadata on the year, country and host species from which the viruses were sampled.

We can visualize these data directly on the phylogeny. In Google Docs, select from the menu File, Download As, Comma-separated values (.csv). This will download the spreadsheet to your Downloads folder.

The Genbank accession numbers approximately match the sequence labels in the phylogeny. To obtain an exact match, we manipulate the annotation file in R:

annot = read.csv("Zika Genomes Jan 2016 -  .csv")
annot2tree = sapply(annot$Genbank.accession,function(acc) grep(acc,ml.tree$tip.label))
annot = cbind("label"=ml.tree$tip.label[annot2tree],annot)
#Save as a tab-deliminted file
write.table(annot,"zika.annotations.txt",row=F,col=T,quote=F,sep="\t")

Return to figtree and open the ML or ClonalFrameML recombination-corrected phylogeny. Open the annotation file by selecting, from the menu, File, Import Annotations and locate the zika.annotations.txt file in your Downloads folder.

Expand the Tip labels panel in the side bar on the left of figtree. There are drop-down menus for selecting the information to Display at the tips of the phylogeny and to Colour by. In addition to the labels, you should now also see Continent, Country, Host and Year, among others. Try experimenting.

Can you answer the following questions? Record your results as you go.

• What was the geographic origin of the 2015/2016 Zika outbreak?
• How was the virus related to previous Zika outbreaks?

There are methods that formally estimate rates of migration between populations, which may be defined for example as countries, continents or host species. If you are interested in these phylogeography methods, I recommend you look up BASTA (De Maio, Wu, O'Reilly, Wilson 2015 PLoS Genetics 11: e1005421) which is implemented in BEAST2.

Calibrating the timescale of Zika virus evolution

There appears to be strong geographic and temporal structuring among the Zika virus sequences. In the following, we are going to focus on the 2015/2016 outbreak virus and its immediate forebears. Returning to R, we will treat HQ234499 as the outgroup and drop all sequences that are more remotely related to the South American outbreak.

subtree.MRCA = getMRCA(ml.tree,c("KU365777.1","HQ234499.1"))
subtree = extract.clade(ml.tree,subtree.MRCA)
#For the next analysis it is necessary to append the tip labels with their sampling dates
subtree$tip.label = paste(subtree$tip.label,annot$Year[match(subtree$tip.label,annot$label)],sep="_")
plot(subtree)
write.tree(subtree,"genomes.subtree.txt")

Is there enough information to estimate the rate of evolution in the Zika virus genomes? For this analysis, it is essential there is no recombination affecting the ancestral history of the sample, which is a good reason to focus on the South American sequences alone.

Open tempest by selecting, from the bottom-left of the screen, Applications, Run Program and typing tempest. A file browser immediately opens: select the genomes.substree.txt file you just created.

Tempest is another program written by Andrew Rambaut with an intuitive graphical interface. Tempest tests for evidence of a statistically significant signal of measurable evolution within the sampling time range of your sequences. Obviously the sequences need to have been sampled over a sufficiently long time period for mutations to have accumulated at a significant amount.

One of the handy features of tempest is its Guess Dates button which automatically 'scrapes' the tip labels for date information. Click the button. Assist tempest by selecting Defined by a prefix and its order, selecting order second and enter an underscore (_) in the prefix box. The sampling years should magically appear when you select OK.

Explore the panels in tempest. You'll see there is a tree panel recapitulating what you plotted in R and a Root-to-tip panel. This takes the form of a scatterplot between the sampling dates of each tip and the phylogenetic distance of each tip from the root. When these cluster closely around the line of best fit, there is likely to be a good signal of measurable evolution.

Take a look at the side bar on the left of tempest. Can you interpret the results to answer the questions:

• When was the most recent common ancestor (MRCA) of the subtree and what does it tell you about Zika virus evolution?
• How quickly is the Zika virus genome evolving, and in what units?
• What is the effect of re-rooting the subtree?

Can you figure out how you would answer:

• When did the 2015/2016 Zika virus strain diverge from other variants?

In the next section we're going to estimate the speed at which Zika virus was spreading using BEAST.