Let's see if Bluesky has some magic ... Is there a way to plot a dated phylogeny with a log-scale time axis? If possible, using R 🧪 #EvolSky #ecoevo @phylogeny.bsky.social #phylo #rstats #plantscience botany - ThreadSky

jgpausas.bsky.social • 22 days ago

Let's see if Bluesky has some magic ...

Is there a way to plot a dated phylogeny with a log-scale time axis? If possible, using R

🧪 #EvolSky #ecoevo @phylogeny.bsky.social #phylo #rstats #plantscience botany

Comments

liamtaylor.bsky.social•22 days ago

If you want to adjust just the breaks/labels on a scale bar, you can use ggtree. With theme_tree2() to show the scale bar. Lets you adjust the x limits/breaks/labels like any other ggplot scale:

ggtree(t) +
scale_x_continuous(limits=c(0, 1000), breaks=c(0, 10, 100, 1000)) +
theme_tree2()

jgpausas.bsky.social•22 days ago

Thanks! That seems to get close but not quite ...

set.seed(11)
tr <- rcoal(9, br=200)
plot(tr); axisPhylo()

ggtree(tr) +
scale_x_continuous(limits=c(0, 2000), breaks=c(0, 10, 100, 1000)) + theme_tree2()

liamtaylor.bsky.social•22 days ago

To flip the axis, you can use revts() on the ggtree object, then set the x axis scale AFTER. You'll need to account for the fact that the x limits/breaks will be - values and use absolute value on the labels.

p <- ggtree(tr) + theme_tree2()
p2 <- revts(p) + scale_x_continuous(labels=abs)

laurenparlett.com•22 days ago

What package are you using to plot? Would it be possible to instead do log(time) and plot that with the phylogeny data?

jgpausas.bsky.social•22 days ago

library(ape)
set.seed(11)
tr <- rcoal(9, br=200)
plot(tr)
axisPhylo()

laurenparlett.com•22 days ago

Hmm... I'm not familiar with that package. I can do some tinkering and see what I can do to help this afternoon.

tayloryoungbauer.bsky.social•22 days ago

Maybe some insight here: https://www.biostars.org/p/266429/#266436

jgpausas.bsky.social•22 days ago

Thanks! But this method disrupts the structure of the phylogeny ... it becomes no ultrametric anymore, etc...

phylogeny.bsky.social•22 days ago

I haven’t tried it myself, but maybe ggtree could be used to format the scale bar?

tlafontrapnouil.bsky.social•22 days ago

Hi. I suppose their is an option in ggtree and associated packages to do so. A workaround might be to directly log your branch length?
Let's say you used ape to import your tree:

mytree$edge.length <- log(mytree$edge.length)

jgpausas.bsky.social•22 days ago

Not that easy!

layout(1:2)
set.seed(11)
tr <- rcoal(9); plot(tr)
tr$edge.length <- log(tr$edge.length)
plot(tr)

tlafontrapnouil.bsky.social•22 days ago

Indeed :D

Got that using log(1+tr$edge.length)
Not quite the same but better

jgpausas.bsky.social•22 days ago

Thank you. Yes, using more realistic values of branch length, the plot is better, but it becomes non-ultrametric

layout(1:2)
set.seed(10)
tr <- rcoal(9, br=200)
plot(tr)
axisPhylo()

tr$edge.length <- log(tr$edge.length)
plot(tr)
axisPhylo()

tlafontrapnouil.bsky.social•22 days ago

Yup, a dirty workaround would be to plot the non ultrametric tree with ggtree. Extract leaves coordinates and add the difference between leaves coordinates and the farthest leaf so that they are all equally far from the root.

tlafontrapnouil.bsky.social•22 days ago

Anyway I feel like you have already better answers.
Hope you'll find out

Comments

Posting Rules

Reply