[Biopython] Phylo: rerooting a tree with a terminal node

[Eric Talevich]

On Wed, Jan 12, 2011 at 9:06 AM, Robert Beiko <beiko at cs.dal.ca> wrote:

> Hi,
>
> I have been experimenting with the excellent Phylo package in BioPython,
> and am having a bit of trouble with the 'root_with_outgroup' method.
>
> Specifically, it seems to work fine when I apply it to internal nodes, but
> when I try to root on a terminal, I get an error:
>
> ---------------
>
> Traceback (most recent call last):
>  File "C:/Projects/10000/Phylogenomics/TestTrees/test.py", line 16, in
> <module>
>    tree.root_with_outgroup(leaf)
>  File "C:\Python26\lib\site-packages\Bio\Phylo\BaseTree.py", line 777, in
> root_with_outgroup
>    parent.clades.pop(parent.clades.index(new_parent))
> ValueError: list.index(x): x not in list
>
> ---------------
>
> Here is an example script:
>
> import io
> import sys
> from Bio import Phylo
>
> infile = 'Example1.tre'
> trees = Phylo.parse(infile,'newick')
>
> for tree in trees:
>    leafList = tree.get_terminals()
>    for leaf in leafList:
>        tree.root_with_outgroup(leaf)
>
> ---------------
>
> And here is the file 'Example1.tre'
>
> ((AAA,BBB),(CCC,DDD));
>
> [I have tried many permutations of the tree, and in no case have I been
> able to root using a terminal].
>
> Line 777 in 'BaseTree.py' is:
> parent.clades.pop(parent.clades.index(new_parent))
>
> so it appears that new_parent is not in the list 'parent.clades'. My Python
> is rather rudimentary so I haven't been able to figure out why this might
> arise.
>
> If I change 'get_terminals()' to 'get_nonterminals()' in the script above,
> everything works fine. I imagine I could get things to work by introducing a
> dummy sister node for each terminal I would like to root on, and then
> rooting on the LCA of the terminal and its dummy sister. But is there
> something I am doing wrong in the script above, that could easily be
> remedied without a hack?
>
> Python version is 2.6, BioPython 1.56.
>
> Best wishes and thanks,
> Rob Beiko
>


Hi Robert,

Thanks for reporting this. It's certainly possible that there's a bug here;
I'll take a closer look at the code.

The odd thing I notice about your code is that you're rerooting inside a
loop. The root_with_outgroup method operates in-place, so the "tree" object
is changing with each iteration. I assume your original code was doing
something after rerooting each time, like writing out a new tree.

The difference in the code between rooting with terminal versus non-terminal
nodes is that rooting with a terminal requires creating a new internal node
just above the terminal, then using the new node as the new root. (Rerooting
with an existing internal node just reroots at that node without creating
any new objects.) So if this is done repeatedly, that could be the source of
some trouble.

Do you know if the loop is crashing in the first iteration or the second?

Best regards,
Eric