Is it possible to assess similarity between two phylo trees?

Question

[shadowstep]

In a recent paper, the authors concluded that the topology of their tree is very similar to a previous study, giving P < 0.001 and ρ = 0.947. I think it could be a correlation test, but I don't know how to achieve this. Can I do this in R? What package/function should I use?

[Joe]

You can look at the Robinson-Foulds metric as an example of comparing topologies.

The ETE3 toolkit is really good for this.

[Chirag Parsania]

Yes, it is possible in the R. Provided that you have same number of elements in both trees, you can convert back phylogenetic tree to distance matrix using R function cophenetic(). Once you get distance matrix, convert it in to object of distance class using function dist(). On the distance class object call function cor() which will give you correlation between two trees.

[h.mon]

What is the paper? It is very unlikely the authors made such claim without specifying which test they performed.

Answer 1

[Frank Wright (BioSS)]

You can carry out tree topology tests easily using IQ-TREE.

See the documentation:http://www.iqtree.org/doc/Command-Reference#tree-topology-tests

One scenario is that you have already estimated Tree(0) from your data and there is an alternative tree Tree(1) in the literature, you supply IQ-TREE with a text file containing the 2 tree files (Newick format). It then carries out a range of significance tests to see if Tree(1) is significantly worse than Tree(0). The tests include Shimodaira-Hasegawa test.

You can easily try this out on th IQ-TREE web server: http://iqtree.cibiv.univie.ac.at/

See "Tree Topology Evaluation and Tests" at bottom of webpage - that is where you upload the file conatining the newick files.