Release announcement: DendroPy 4.3.0

[Jeet Sukumaran]

DendroPy 4.3.0 has been released!

Folks, this one includes a VERY important bugfix for SumTrees, so I highly recommend everyone upgrade to this.

Even if you do not use SumTrees, you probably want to grab this for all the goodies that it offers. Apart from the usualy minor tweaks and fixes, you are going to be getting:

-   A new ``sumlabels.py`` application: (string) concatenation of labels for corresponding edges across multiple trees.

-   The birth-death tree simulation (``dendropy.model.birth_death_tree``) now allows for preservation of extinct tips.

-   Improved performance of character subset export.

For more information:

    http://dendropy.org

Get it now with:

    $ sudo pip install --upgrade dendropy

or:

    $ sudo pip install -U dendropy

You can install directly from the main GitHub repository using:

    $ pip install git+https://github.com/jeetsukumaran/DendroPy.git

or:

    $ pip install git+git://github.com/jeetsukumaran/DendroPy.git

Or, if you are working out of a local Git repo:

    $ git pull origin

    $ python setup.py build

    $ sudo python setup.py develop

(note: ``sudo`` should be used only if you are installing system-wide using your system Python).

[Jeremy Brown]

Hi Jeet,

You put the fear in us, so my lab is updating all of our installations to 4.3.0. Can you tell us what the bug was in SumTrees? 

Thanks,

Jeremy

--
You received this message because you are subscribed to the Google Groups "DendroPy Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dendropy-user...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Jeet Sukumaran]

Hi Jeremy,

Ah, yes, I guess I should have provided that detail in the announcement,
though it is there in the change log.

Basically, there was a corner case in which (a) if you asked for a
consensus tree to be constructed and have support mapped onto that, and
(b) your source trees were unrooted but you asked for them to be treated
as rooted (using the ``--rooted'' flag), and (c) the input trees were in
NEWICK format, there might be a split on the consensus tree that would
incorrectly have 0 support and 0 branch length. Basically, the reason
was that the consensus tree split hashes were calculated at the wrong
time (after some processing, instead of before), and one split might be
be assigned the wrong hash.

The 0-support for a bifurcation on the consensus tree is an almost sure
indication of the problem: the consensus tree is constructed for the
true/correct support values, so if a split appears on the consensus tree
than the support value is (by default) at least > 0.5; so if it later
has 0 support mapped onto it, then you can take it as a red flag.

Hope you were not affected by this! Apologies if you were, and let me
know if I can help in any way or if you encounter other problems.

-- jeet

On 7/11/17 12:51 PM, Jeremy Brown wrote:
> Hi Jeet,
>
> You put the fear in us, so my lab is updating all of our installations
> to 4.3.0. Can you tell us what the bug was in SumTrees?
>
> Thanks,
> Jeremy
>
>
>
>> On Jul 6, 2017, at 8:37 PM, Jeet Sukumaran <jeetsu...@gmail.com

>> <mailto:dendropy-user...@googlegroups.com>.

>> For more options, visit https://groups.google.com/d/optout.
>

> --
> You received this message because you are subscribed to the Google
> Groups "DendroPy Users" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to dendropy-user...@googlegroups.com

> <mailto:dendropy-user...@googlegroups.com>.

> For more options, visit https://groups.google.com/d/optout.

--



--------------------------------------
Jeet Sukumaran
--------------------------------------
jeetsu...@gmail.com
--------------------------------------
Blog/Personal Pages:
http://jeetworks.org/
GitHub Repositories:
http://github.com/jeetsukumaran
Photographs (as stream):
http://www.flickr.com/photos/jeetsukumaran/
Photographs (by galleries):
http://www.flickr.com/photos/jeetsukumaran/sets/
--------------------------------------

[Jeremy Brown]

Hi Jeet,

Thanks! I looked at the change history page on the website and didn’t see it, but now I notice it’s on GitHub. That is certainly a corner case and didn’t affect us. A 0-support label on a consensus tree would hopefully be spotted right away, in any case.

Thanks again for such a great library. We use it all the time.

Jeremy