1
$\begingroup$

I'm interested know about some basic metrics that could be used to discriminate trees based on their branch lengths. I know of branch length unaware topological/tree shape-based methods such as Colless and Sackin's indices and am curious whether there are similar metrics that utilize branch length to discriminate. For example, the two trees below have an identical topology but are very different.

Tree 1

Tree 2

$\endgroup$
  • 1
    $\begingroup$ Welcome to Biology.SE! Please take the tour and then go through the help pages starting with How to Ask questions effectively on this site. In particular, this question seems like it might be a better fit for Bioinformatics. (Note: please don't repost your question to that site — instead request migration). Thanks! 😊 $\endgroup$ – tyersome Jun 11 at 23:11
  • $\begingroup$ Thanks @tyersome! This similar post elsewhere on this site brought me here. $\endgroup$ – Floris Jun 12 at 2:14
  • 1
    $\begingroup$ You're welcome! Bioinformatics.SE didn't exist 4 years ago, so that isn't really a good indicator of what is currently a good fit for this site ... $\endgroup$ – tyersome Jun 12 at 2:42
  • $\begingroup$ Are you essentially asking how to interpret the difference in the two phylo trees(that is,given the tree,what do you conclude),or are you asking which tree to produce for your data(that is,given the items,which tree to produce)?The horizontal length gives you a sense that two things are very separate or not.Looking at the second tree,you could say that right at the beginning of time,d,e,f differentiated but a,b,c remained identical until some much later date when they finally split and differentiated.The horizontal distance doesn't have to be time-any measure that measures similarity will do. $\endgroup$ – JimN Aug 26 at 7:24
0
$\begingroup$

I don't know whether this is actually used in phylogeny, but if you restrict the problem to cases where the topologies are the same, each tree can be described as a point in a n-dimensional space where n is the number of branches, and each axis represents the (possibly normalized) length of one of the branches.

You can then compute distances between points (i.e. trees) in that space.

| improve this answer | |
$\endgroup$
0
$\begingroup$

There are a variety of measures of tree similarity implemented in the TreeDist R package. (Your phylogenies appear to be from the R ape package, so hopefully you are somewhat familiar with R.)

Other tools include treedist from Joe Felsenstein, which computes Robinson-Foulds and the Branch Score Distance (which sounds closely related to your interests, described in this paper). You can see that software page for a lot more information.

For a practical description of the different tree comparison metrics you can see this article.

I would suggest taking a look at the manual and the multiple vignettes associated with the TreeDist package to learn more about how exactly those metrics work. There appear to be a wide variety of metrics in there that I have never heard of, such as clustering entropy or split entropy or Nye similarity.

This package also includes tools for making visualizations of differences, such as that shown below:

enter image description here

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.