RAxML-NG 1.0.1 bootstrap: "terminate called recursively"

[Nicolás Peñafiel]

Dear community:

I have been having an issue when running bootstrap for my tree. 

After starting, at some, sooner or later the job ends with the message: "terminate called recursively" in the scheduler's log file.

The code I used is:

#SBATCH --cpus-per-task=8

#SBATCH --mem=1G

module load StdEnv/2020  gcc/9.3.0  openmpi/4.0.3  raxml-ng/1.0.1

export OMP_NUM_THREADS="${SLURM_CPUS_PER_TASK:-1}"

raxml-ng --bootstrap --msa TVM.raxml.rba --model TVM+I+G4  --seed 5 --threads ${OMP_NUM_THREADS} --prefix TVM_BS_2 --bs-trees 5000

I had previously determined the best model using Modeltest. The job doesn't end due to a lack of time or memory. 

I am attaching the raxml's log file, the scheduler log file and the MSA.

Your help is appreciated.

Nicolas

[Nicolás Peñafiel]

Hi again. 

I tried running the bootstrap for that MSA again (with a different seed and prefix) and again it stopped after 56 replicates, only that a new error message was thrown:

ERROR: ERROR in SPR round (LIBPLL-2240): BL opt converged to a worse likelihood score by -546.001568175968714 units

This looks like the error is indeed related to RAxML rather than to the scheduler, which I thought of as one of the possible causes.

Any insight about what is the underlying cause of that error?

thanks in advance

[Alexey Kozlov]

Hello Nicolás,

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

Thanks,
Alexey

> --
> You received this message because you are subscribed to the Google Groups "raxml" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to
> raxml+un...@googlegroups.com <mailto:raxml+un...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/raxml/76eb9e65-16e2-4ad4-9baa-20661221a9c2n%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/76eb9e65-16e2-4ad4-9baa-20661221a9c2n%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Nicolás Peñafiel]

Hi Alexey, thanks for looking into it.

Here is the code I used:

#!/bin/bash

#SBATCH --job-name=boots_TVM_3

#SBATCH --time=0-3:00:00

#SBATCH --cpus-per-task=8

#SBATCH --mem=1G

module load StdEnv/2020  gcc/9.3.0  openmpi/4.0.3  raxml-ng/1.0.1

export OMP_NUM_THREADS="${SLURM_CPUS_PER_TASK:-1}"

raxml-ng --bootstrap --msa TVM.raxml.rba --model TVM+I+G4  --seed 11 --threads ${OMP_NUM_THREADS} --prefix TVM_BS_3 --bs-trees 5000

Here are the documents:

- The MSA alignment

-The log file from RAxML-NG

- The log file from the scheduler

cheers

Nicolas

[Alexey Kozlov]

Hi Nicolas,

thanks for sending the files!

I can confirm there is a problem, and it looks very similar to the one which has been already
reported, which I am however still investigating and do not have a final fix yet.

In the meantime, you can use a simple workaround, just add

--blopt nr_safe

option to the command line.

As a side note, I would recommend reducing the number of bootstrap replicates, you can use e.g.

--bs-tree autoMRE{5000}

to perform automatic bootstrap convergence test.

Best,
Alexey

On 29.09.21 01:51, Nicolás Peñafiel wrote:
> Hi Alexey, thanks for looking into it.
> Here is the code I used:
>
> #!/bin/bash
>
> #SBATCH --job-name=boots_TVM_3
>
> #SBATCH --time=0-3:00:00
>
> #SBATCH --cpus-per-task=8
>
> #SBATCH --mem=1G
>
> module load StdEnv/2020  gcc/9.3.0  openmpi/4.0.3  raxml-ng/1.0.1
>

> export OMP_NUM_THREADS=*"${SLURM_CPUS_PER_TASK:-1}"*
>
> raxml-ng*--bootstrap --msa* TVM.raxml.rba*--model* TVM+I+G4 *--seed* 11*--threads*
> *${OMP_NUM_THREADS}**--prefix* TVM_BS_3*--bs-trees* 5000

> <https://groups.google.com/d/msgid/raxml/76eb9e65-16e2-4ad4-9baa-20661221a9c2n%40googlegroups.com?utm_medium=email&utm_source=footer
> <https://groups.google.com/d/msgid/raxml/76eb9e65-16e2-4ad4-9baa-20661221a9c2n%40googlegroups.com?utm_medium=email&utm_source=footer>>.

>
>
> --
> You received this message because you are subscribed to the Google Groups "raxml" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to
> raxml+un...@googlegroups.com <mailto:raxml+un...@googlegroups.com>.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/raxml/30f9bc66-fe62-4f9f-9a79-8cf2a2826770n%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/30f9bc66-fe62-4f9f-9a79-8cf2a2826770n%40googlegroups.com?utm_medium=email&utm_source=footer>.