Rooting phylogenetic trees

[Carly]

Hi,

From the make_phylogeny.py the default setting is using Fasttree tree building method.  I try to import my .tre file into R using the phyloseq package and it indicates that the tree is not rooted. I asked this on phyloseq github issues and Paul indicated that he did not know why my tree isn't rooted. I was hoping you all could clarify this for me. 

I saw on this post on 10/17/12 https://groups.google.com/forum/#!searchin/qiime-forum/root$20a$20tree/qiime-forum/XvPErZ6e-fI/JXy3u4BRvkgJ 

about using an archael outgroup to root the tree.

I was curious to your all's opinion on rooting microbial phylogenies and how this might affect downstream analyses that use the .tre file. I would assume that using a random mid-point would be the worst as it assumes a molecular clock which seems to be controversial (http://guava.physics.uiuc.edu/~nigel/courses/598BIO/498BIOonline-essays/hw2/files/hw2_li.pdf).

Otherwise, it is using an outgroup or the Fasttree rooting method that is the default of make_phylogeny.py. But now I'm wondering if this is only building the tree and not rooting it.

Anyway, any help would be greatly appreciated.

Cheers,

Carly

[Daniel McDonald]

Hey Carly,

The trees produced by FastTree are not rooted but make_phylogeny.py
optionally allows you to perform midpoint rooting. One challenge here
is that, as you noted, it is not clear what the root should be on some
trees. Blindly adding in an archaeal outgroup may be unreasonable if
there is to much divergence between the outgroup and the leaves of the
tree. AFAIK, no one has benchmarked the affects of using rooted or
unrooted trees within QIIME.

QIIME handles rooted trees (e.g., the Greengenes tree) the same as
unrooted trees, so if you root the tree using an external application
following make_phylogeny.py, it can still be used for downstream
methods.

Best,
Daniel

> --
>
> ---
> You received this message because you are subscribed to the Google Groups
> "Qiime Forum" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to qiime-forum...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>

[Mike]

I have some experience with observing quite different results when alternating between using outgroups. I found that the most reliable solution for my data was to build a phylogeny with Archaea as the outgroup taxa. To ensure that the tree was rooted I opened the tree file in Dendroscope and manually rooted the entire phylogeny to the Archaea (by looking up the OTU identifiers in my taxonomy file).

Now, there is a long branch from the Archaea to Bacteria which can alter the UniFrac or any other phylogneetic community metric. So, after I root the tree I use Dendroscope again to prune the Archaea outgroup (and the branch leading to them). This way I have a properly constructed and rooted tree based on the outgroup, but without the side-effect of incorporating the long branch to Archaea (which would affect the branchlength calculations). 

I have taken this approach when using QIIME to process invertebrate data using the cox1 gene. Here, the closest outgroup had a substantially long branch and contributed to significantly different values of UniFrac distances as compared to not using the outgroup. Since, I was interested in differences in communities and not how the differences in communities compared to the origin of the group in question, I rooted the tree with the outgroup and then pruned that outgroup from the tree in Dendroscope. This way I had a properly rooted tree without the effect of biases caused by the long branches leading to the outgroup. Then I performed all my phylogenetic community analysis.

So, it comes down to your question. If you only care about telling the differences between communities, then you should always prune to the closest outgroup. Otherwise if your question is "how different are these communities with respect to the tree of life, or the history of this group" then you may want to retain the outgroup. I always like to think about the question from a phylogenetic standpoint and then determine if I should retain the outgroup or not. This is important, as you should think of the phylogenetic scale you are working with. Again, these choices depend on the phylogenetic question you want to answer.

Anyway, those are my 2 cents. I'd love to hear the thoughts of others on this.

-Mike