BL opt converged to a worse likelihood score

[jonoave]

Hello,

I just encountered an error running raxml-ng:

ERROR: ERROR in branch length optimization (LIBPLL-2240): BL opt converged to a worse likelihood score by -0.016206621156243 units terminate called after throwing an instance of 'std::runtime_error'

The raxml version is:
RAxML-NG v. 0.9.0git released on 24.05.2019 by The Exelixis Lab.

I found a similar error referenced here #48, but it was stated as solved.

The command was:
raxml-ng --model LG4X+IC --msa KOG0003.trim.gap.min --threads 2

Edit to add:
Changed model to LG4M+IC, it ran without any issues.

[amkozlov]

Hi @jonoave,

could you please send me your input files and raxml-ng log?

[jonoave]

Hi,

Here are the files
KOG0003.trim.gap.min.raxml.log

KOG0003.trim.gap.txt

[amkozlov]

Hi @jonoave,

thanks for reporting and sending the input files!

This bug is now fixed in the dev branch.

[jbruxaux]

Hi,

I'm having exactly the same problem with version 1.1.0. Most of my gene trees run fine, but two of them fail. I attach one (phy format but renamed as txt), in case it helps. Is it a matter of not enough informative sites inside?

Error message:

terminate called after throwing an instance of 'std::runtime_error'
  what():  ERROR in SPR round (LIBPLL-2240): BL opt converged to a worse likelihood score by -0.080800135211916 units
/var/spool/slurmd/job20331448/slurm_script: line 35: 665385 Aborted                 (core dumped) raxml-ng --threads 16 --bootstrap --bs-trees autoMRE{10000} --bs-cutoff 0.01 --msa /pfs/stor10/projects/projnb10/snic2019-29-3/chloroplast_data/gene_trees/${file}_T2.raxml.rba --model GTR+G --redo --prefix /pfs/stor10/projects/projnb10/snic2019-29-3/chloroplast_data/gene_trees/${file}_T4

psaI.txt

Thanks for your help!

Jade

[amkozlov]

Hi @jbruxaux

can you please attach full log of this run?

[jbruxaux]

Hi @amkozlov

Here it is!
Thanks for your help!

raxml_psaI.txt

[amkozlov]

thanks, please try with --blopt nr_safe, it should help.

[jbruxaux]

Thanks @amkozlov it worked fine with this option!
You can close the issue.