decomposition problem on unresolved edges

[smirarab]

If the tree used to decompose the set of taxa has polytomies with degree larger than maximum subset size, sate divides the poltyomy into two subsets, one with size 1 and the other with side d-1 (where d is the degree of the ploytomy).

If the tree and the polytomy are both very large, this can result in maximum recursion error.

The easiest solution randomly resolving polytomies for any tree inside SATe. This can be achieved easily by adding the following line:

        dendropy_tree.resolve_polytomies(update_splits=True)

as the first line of the constructor of PhylogeneticTree (in tree.py line 24).

This is not a imaginary case. FastTree outputs polotyomies when input sequences are identical. And we have faced this problem on a bunch of real biological datasets we are testing.

If this solution is fine with everyone, I can go ahead and apply it to the code.

Thanks
Siavash

[mtholder]

that sounds fine to me.

Mark

On Apr 16, 2012, at 4:17 PM, Siavash Mirarab wrote:

If the tree used to decompose the set of taxa has polytomies with degree larger than maximum subset size, sate divides the poltyomy into two subsets, one with size 1 and the other with side d-1 (where d is the degree of the ploytomy).

If the tree and the polytomy are both very large, this can result in maximum recursion error.

The easiest solution randomly resolving polytomies for any tree inside SATe. This can be achieved easily by adding the following line:

       dendropy_tree.resolve_polytomies(update_splits=True)

as the first line of the constructor of PhylogeneticTree (in tree.py line 24).

This is not a imaginary case. FastTree outputs polotyomies when input sequences are identical. And we have faced this problem on a bunch of real biological datasets we are testing.

If this solution is fine with everyone, I can go ahead and apply it to the code.

Thanks
Siavash

---

Reply to this email directly or view it on GitHub:
#13

[mtholder]

Hi Siavash,
I decided that it would be a little odd for a class called "PhylogeneticTree" to modify the tree on initialization. So I added a force_fully_resolved attribute, and I used the resolve_polytomies call in a couple of places (corresponding to when new trees are read).

This should be fixed after commit 73f1d96 on 26 Apr, 2012.
Can you verify that this fix works on a dataset that has this problem (or send the dataset to me or Jamie)?

Thanks

[smirarab]

HI Mark,

Sure, I will test this.

Regards
Siavash

On Thu, Apr 26, 2012 at 2:26 PM, Mark Holder <
[email protected]

wrote:

Hi Siavash,
I decided that it would be a little odd for a class called
"PhylogeneticTree" to modify the tree on initialization. So I added a
force_fully_resolved attribute, and I used the resolve_polytomies call in a
couple of places (corresponding to when new trees are read).

This should be fixed after commit 73f1d96
on 26 Apr, 2012.
Can you verify that this fix works on a dataset that has this problem (or
send the dataset to me or Jamie)?

Thanks

---

Reply to this email directly or view it on GitHub:
#13 (comment)

[mtholder]

Actually hold off for a minute I have a commit that I'm about to push ...
On Apr 26, 2012, at 2:42 PM, Siavash Mirarab wrote:

HI Mark,

Sure, I will test this.

Regards
Siavash

On Thu, Apr 26, 2012 at 2:26 PM, Mark Holder <
[email protected]

wrote:

Hi Siavash,
I decided that it would be a little odd for a class called
"PhylogeneticTree" to modify the tree on initialization. So I added a
force_fully_resolved attribute, and I used the resolve_polytomies call in a
couple of places (corresponding to when new trees are read).

This should be fixed after commit 73f1d96
on 26 Apr, 2012.
Can you verify that this fix works on a dataset that has this problem (or
send the dataset to me or Jamie)?

Thanks

---

Reply to this email directly or view it on GitHub:
#13 (comment)

---

Reply to this email directly or view it on GitHub:
#13 (comment)

[mtholder]

oops. bug fix. try with 81bb219 or later (I have to bundle dendropy's resolve_polytomies because the GUI bundle has an old dendropy)