generating bootstraps with fixed topology, running into error

[Pascal Title]

Hi,

I would like to generate bootstrap replicates on a fixed topology in order to get branch length variation.

For setup context, I'm running raxml-ng v1.2.1-master installed via Homebrew, on a M2 Apple Macbook pro.

Based on information I've seen on the wiki and in the google group, I've done the following:

Generate bootstrap replicates of alignment with:
raxml-ng --bsmsa --msa ../concat.phy --model ../raxml_magusAln.best_scheme.txt --prefix BOOT --bs-trees 100 --seed 12345 

This generated 100 bootstrap alignments. Then for each of these 100 alignments, do:

raxml-ng --evaluate --msa BOOT.raxml.bootstrapMSA.1.phy --model BOOT.raxml.bootstrapMSA.partition --tree ../raxml_alldat.raxml.bestTree --prefix EVAL1 --threads 10

Trying this on the first bootstrap replicate, I ran in to the following error:

[00:00:00 -540715.323541] Initial branch length optimization
[00:00:01 -517974.570157] Model parameter optimization (eps = 10.000000)
libc++abi: libc++abi: terminating due to uncaught exception of type std::runtime_error: ERROR in branch length optimization (LIBPLL-2240): BL opt converged to a worse likelihood score by -1031.857464378757868 unitsterminating due to uncaught exception of type std::runtime_error: ERROR in branch length optimization (LIBPLL-2240): BL opt converged to a worse likelihood score by -1031.857464378757868 units
libc++abi: libc++abi: libc++abi: [1]    54106 abort      raxml-ng --evaluate --msa BOOT.raxml.bootstrapMSA.1.phy --model  --tree     1

I then added the flag --blopt nr_safe as suggested on another thread here, but that leads to the following error: 

[00:00:00 -540715.323541] Initial branch length optimization
[00:00:02 -517974.534626] Model parameter optimization (eps = 10.000000)
Assertion failed: (new_loglikelihood - loglikelihood > new_loglikelihood * BETTER_LL_TRESHOLD), function pllmod_opt_optimize_branch_lengths_local_multi, file pll_optimize.c, line 1903.
[1]    54166 abort      raxml-ng --evaluate --msa BOOT.raxml.bootstrapMSA.1.phy --model  --tree     1

Any thoughts or suggestions?

Thanks!

-Pascal

[Oleksiy Kozlov]

Hi Pascal,

as a quick workaround, please try to add "--simd none", or run in Rosetta (x86 emulation mode).

If you send me your input data, I'll try to find an M2 machine and investigate it further.

Best,
Oleksiy

> --
> 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/50f69ff4-7cd5-437b-877f-a51d8c159fffn%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/50f69ff4-7cd5-437b-877f-a51d8c159fffn%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Pascal Title]

I tried with the additional --simd flag, but got the following errors.

Evaluating 1 trees

[00:00:00] Tree #1, initial LogLikelihood: -540715.323541

[00:00:00 -540715.323541] Initial branch length optimization
[00:00:01 -517974.570157] Model parameter optimization (eps = 10.000000)

libc++abi: terminating due to uncaught exception of type std::runtime_error: ERROR in branch length optimization (LIBPLL-2240): BL opt converged to a worse likelihood score by -20.770657733024564 units
libc++abi: terminating due to uncaught exception of type std::runtime_error: ERROR in branch length optimization (LIBPLL-2240): BL opt converged to a worse likelihood score by -20.770657733024564 units
libc++abi: terminating due to uncaught exception of type std::runtime_error: ERROR in branch length optimization (LIBPLL-2240): BL opt converged to a worse likelihood score by -20.770657733024564 units[1]    59155 abort      raxml-ng --evaluate --msa BOOT.raxml.bootstrapMSA.1.phy --model  --tree     1

I did switch to a linux machine, and everything is working fine there. So from my perspective, I'm all set. But if you would like my input files in order to sort out the mac related issue, please let me know (and where to send) and I'd be happy to share the input files.

Thanks for your help!

-Pascal

[Oleksiy Kozlov]

Perfect, if possible please send the input files

BOOT.raxml.bootstrapMSA.1.phy
BOOT.raxml.bootstrapMSA.partition
raxml_alldat.raxml.bestTree

to my e-mail address.

Thank you!

> <https://groups.google.com/d/msgid/raxml/50f69ff4-7cd5-437b-877f-a51d8c159fffn%40googlegroups.com?utm_medium=email&utm_source=footer <https://groups.google.com/d/msgid/raxml/50f69ff4-7cd5-437b-877f-a51d8c159fffn%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/14192ec5-3fd4-4fc7-839f-463c639ef269n%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/14192ec5-3fd4-4fc7-839f-463c639ef269n%40googlegroups.com?utm_medium=email&utm_source=footer>.