RAxML-EPA: visualisation with Archaeopteryx

[Fabien Burki]

Hello,

I ran a RAxML-EPA analysis and am trying to visualise the resulting .jplace file with archaeopteryx. For that purpose, I first converted the .jplace file into a phyloxml format using guppy in combination with the tog command:

guppy tog --xml --out-dir outdir infile.jplace

I can now open the .xml file with Archaeopteryx, and visualise the branch colours. The issue I have is that all branches leading to the placed reads are red (see attached).

My questions are:

1) Are the colours representing the likelihood weight as defined in the RAxML_classificationLikelihoodWeights file, or the EDPL measurement of pplacer?

2) How are the colours defined, i.e. what does red (and other colours) represent?

3) I do not have a good sense of what is considered a robust placement or not, I would really like to have your opinion on this.

Thank you so much for your help,

Fabien

[Alexey Kozlov]

Dear Fabien,

> 1) Are the colours representing the likelihood weight as defined in
> the RAxML_classificationLikelihoodWeights file, or the EDPL
> measurement of pplacer?
>
> 2) How are the colours defined, i.e. what does red (and other colours)
> represent?
>

My guess is that all branches leading to the query sequences are just
marked red, irrespective of their LH weights. At least, we do not
provide any color coding in .jplace output file you get from RAxML-EPA.
So if there is any, it must be introduced by guppy tog and/or
Archaeopteryx, hence I suggest that you refer to the respective manuals
or search advice from their developers.

> 3) I do not have a good sense of what is considered a robust placement
> or not, I would really like to have your opinion on this.
>

If you get a placement to a certain branch (or to the several adjacent
branches) with the high likelihood weight, I'd consider this placement
to be "robust". On the contrary, if placements are "scattered" across
the tree, with no obvious best-scoring branch(es), that could be an
indication of lack of clear phylogenetic signal in the data.

You can find the likelihood weights for each read and branch in the
.jplace file.

Cheers,
Alexey