How To Get Bootstrap Values From A Tree Generated By The Consense Program (Phylip)?

Question

[Panos]

I simply want to create a 16S-based phylogenetic tree. I first used seqboot (1000 replicates), then dnadist, then neighbor, then consense and finally retree (change outgroup).

By searching the internet I realized that consense "confuses" branch length with bootstrap values... I opened the tree file using a text editor and it looks like this:

(Haloarc_am:500.0,(((((S000128561:1000.0,(((((KK2_8:1000.0,S000372751:1000.0):1000.0,

S000385187:1000.0):515.0,((S000434611:1000.0,S000120839:1000.0):537.0,KK37:1000.0):1000.0):1000.0, (KK210:1000.0,S001177649:1000.0):1000.0):660.0,(KK154:1000.0,S001801122:1000.0):1000.0):569.0):775.0, (S000126695:1000.0,KK311:1000.0):1000.0):978.0,(((((S000103513:1000.0,

What I can't understand is which of the numbers are bootstrap values; I suspect some of them are branch lengths (maybe those exactly next to taxa names).

[Panos]

I guess I should've seen it earlier that there are only bootstrap values in here and no branch lengths... Even the scale bar is completely wrong! So, what to do if I want to create a phylogenetic tree with branch lengths AND bootstrap values on it?

Answer 1

[Michael Kuhn]

It looks like all of the numbers (which are all integers) are the number of trees in which the sub-tree was found. Since you did 1000 bootstraps, 775.0 should correspond to a bootstrap support of 77.5%.

[David W]

Yup. There are no branch lengths on a consense tree - just the topology

Answer 2

[aidan-budd]

What you need to do here is to take the topology of the consensus tree you've obtained from CONSENSE, and then provide this topology, together with your aligment, to software which will calculate branch lengths for that topology in the context of that alignment using some model. RAxML would do that for you, I think also PhyML, PUZZLE, and I think PROML (in the phylip package).

Then you need to associated the lengths of the branches (which are the "bootstrap values" you've calculated) of the tree you got out of CONSENSE with the (probably calculated using maximum likelihood) topology + branch lengths you've calculated using e.g. RAxML