Unexpected unifurcation suppression in reroot_at_node()

[psathyrella]

Unless I'm missing something, it appears that when reroot_at_node() is called on an unrooted tree, that suppress_unifurcations=False doesn't always have the desired effect. I think this isn't the expected behavior?

This should reproduce it:

#!/usr/bin/env python
import dendropy

print 'version: ', dendropy.__version__
unrooted_treestr = '(((l1,l2)n2)n1)root;'
rooted_treestr = '[&R] ' + unrooted_treestr

for tlabel, tstr in [('unrooted', unrooted_treestr), ('rooted', rooted_treestr)]:
    print '%s:' % tlabel
    dtree = dendropy.Tree.get_from_string(tstr, 'newick')

    print 'before reroot:'
    print dtree.as_ascii_plot(show_internal_node_labels=True, width=70)

    new_root_node = dtree.seed_node.child_nodes()[0]
    dtree.reroot_at_node(new_root_node, suppress_unifurcations=False)

    print 'after reroot:'
    print dtree.as_ascii_plot(show_internal_node_labels=True, width=70)
    print ''

and the output I get is:

version:  4.4.0
unrooted:
before reroot:
                                            /---------------------- l1
root------------------n1--------------------n2                        
                                            \---------------------- l2

after reroot:
/---------------------------------------------------------------- l1  
|                                                                     
n1--------------------------------------------------------------- l2  
|                                                                     
\---------------------------------------------------------------- root

rooted:
before reroot:
                                            /---------------------- l1
root------------------n1--------------------n2                        
                                            \---------------------- l2

after reroot:
                                /-------------------------------- l1  
/-------------------------------n2                                    
n1                              \-------------------------------- l2  
|                                                                     
\---------------------------------------------------------------- root

where n2 disappears on reroot in the first (unrooted) case, but not in the second (rooted) case.

[jeetsukumaran]

What's happening here is that because the first tree is unrooted, any basal bifurcation that results will be collapsed by default. Most of the lower level tree manipulation functions take a collapse_unrooted_basal_bifurcation argument to prevent this from happening (by setting this to False). Unfortunately, in higher-level functions such as reroot_at_node, I have not been diligent about allowing this parameter to be set to be passed down to the lower level functions.

I have just done so in the most recent revision here: 86d66160 .

Note that a basal bifurcation on an unrooted tree results in a splits hash collision --- the two child edges of the root note will have the exact same identity, and in the dictionary lookup only one of these will be stored. I would suggest sticking to rooted trees if the basal split is important to you, as on an unrooted tree this is all an illusion.

On 7/3/19 2:15 PM, Duncan Ralph wrote:

Unless I'm missing something, it appears that when reroot_at_node() is called on an unrooted tree, that |suppress_unifurcations=False| doesn't always have the desired effect. I think this isn't the expected behavior?

This should reproduce it:

|#!/usr/bin/env python import dendropy print 'version: ', dendropy.version unrooted_treestr = '(((l1,l2)n2)n1)root;' rooted_treestr = '[&R] ' + unrooted_treestr for tlabel, tstr in

'%s:' % tlabel dtree = dendropy.Tree.get_from_string(tstr, 'newick') print 'before reroot:' print dtree.as_ascii_plot(show_internal_node_labels=True, width=70) new_root_node = dtree.seed_node.child_nodes()[0] dtree.reroot_at_node(new_root_node, suppress_unifurcations=False) print 'after reroot:' print dtree.as_ascii_plot(show_internal_node_labels=True, width=70) print '' |

and the output I get is:

|version: 4.4.0 unrooted: before reroot: /---------------------- l1 root------------------n1--------------------n2 ---------------------- l2 after reroot: /---------------------------------------------------------------- l1 | n1--------------------------------------------------------------- l2 | ---------------------------------------------------------------- root rooted: before reroot: /---------------------- l1 root------------------n1--------------------n2 ---------------------- l2 after reroot: /-------------------------------- l1 /-------------------------------n2 n1 -------------------------------- l2 | ---------------------------------------------------------------- root |

where |n2| disappears on reroot in the first (unrooted) case, but not in the second (rooted) case.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/jeetsukumaran/DendroPy/issues/118?email_source=notifications&email_token=AAAGMRYWGGX4DNNXZUHDQD3P5UJFVA5CNFSM4H5LAR5KYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4G5HNHFA, or mute the thread https://github.com/notifications/unsubscribe-auth/AAAGMRZEGN7XSUSATP4BVMLP5UJFVANCNFSM4H5LAR5A.

--

---

Jeet Sukumaran

Assistant Professor Biology Department San Diego State University

Lab: https://sukumaranlab.org/ Blog: https://jeetblogs.org/ Repositories: https://github.com/jeetsukumaran Photography: https://www.flickr.com/photos/jeetsukumaran/ Instagram: https://www.instagram.com/jeetsukumaran/ Calendar: https://goo.gl/dG5Axs

Email: jsukumaran@sdsu.edu (work) jeetsukumaran@gmail.com (personal)

Mailing Address: Biology Department, LS 262 San Diego State University 5500 Campanile Drive San Diego, CA 92182-4614

[psathyrella]

ok, great, thanks.

Also, I may have missed it, but I couldn't find the choices for the rooting keyword arg for Tree.get() by clicking around from here or here, although eventually found it in dataio/newickreader.py by grepping the source.