Segfault when building a consensus tree

[BenoitMorel]

Hi Diego,

when running ./modeltest-ng -i ENSGT00760000118962.fa -t mp

I get a segfault

#3  0x00000000005c399e in __assert_fail ()
#4  0x00000000004b08f2 in build_tips_recurse (tree=0x2a64c40, tip_labels=0x1e26020, split_len=6) at consensus.c:927
#5  0x00000000004b0662 in build_tips_recurse (split_len=6, tip_labels=0x1e26020, tree=0x1fb39a0) at consensus.c:946
#6  build_tips_recurse (split_len=6, tip_labels=0x1e26020, tree=0x1fb2fe0) at consensus.c:946

You can find the fasta file at https://github.com/BenoitMorel/phd_experiments/blob/master/datasets/bugs/modeltest/ENSGT00760000118962.fa

I reproduced the issue when building from the most recent sources, but also with the last binary release.

Benoit

[dapogon]

This alignment seems to have empty columns... --

On Wed, Apr 11, 2018 at 11:42 AM BenoitMorel notifications@github.com wrote:

Hi Diego,

when running ./modeltest-ng -i ENSGT00760000118962.fa -t mp

I get a segfault

3 0x00000000005c399e in __assert_fail ()

4 0x00000000004b08f2 in build_tips_recurse (tree=0x2a64c40, tip_labels=0x1e26020, split_len=6) at consensus.c:927

5 0x00000000004b0662 in build_tips_recurse (split_len=6, tip_labels=0x1e26020, tree=0x1fb39a0) at consensus.c:946

6 build_tips_recurse (split_len=6, tip_labels=0x1e26020, tree=0x1fb2fe0) at consensus.c:946

You can find the fasta file at https://github.com/BenoitMorel/phd_experiments/blob/master/datasets/bugs/modeltest/ENSGT00760000118962.fa

I reproduced the issue when building from the most recent sources, but also with the last binary release.

Benoit

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ddarriba/modeltest/issues/14, or mute the thread https://github.com/notifications/unsubscribe-auth/ATlSEn7MMYEny25229pIuTgVM7JBCACyks5tndAHgaJpZM4TPsxN .

-- David Posada Facultad de Biología Campus Universitario Universidad de Vigo 36310 Vigo Spain

Phone: +34 986 812038 Cell: +34 647 343300 Fax: +34 986 812556 Email: dposada@uvigo.es Web: http://darwin.uvigo.es

[ddarriba]

Thank you Benoit!

This error comes from a bug in pll-modules, that occurs when the number of taxa is a multiple of 32. I'll fix it, and please update pll-modules in your software if they use split-related functions.

Best, Diego.

[BenoitMorel]

Dear David and Diego, thanks a lot for your quick answers. I will temporary exclude multiple of 32 taxa datasets from my analysis. In general, should I filter out empty columns from my MSAs myself before running modeltest? Best, Benoit

[ddarriba]

Thank you for the report ;)

I will update pll-modules in modeltest later this afternoon, but you can already link manually to the latest modules commit and use your multiple-of-32-taxa MSA. Empty columns are not a problem.

Best, Diego.

[BenoitMorel]

It works fine now, thanks a lot!