Get bootstrap values on consensus tree

Pailler Vincent

unread,

Apr 2, 2024, 6:16:14 AMApr 2

to raxml

Dear Oleksiy,

I work on a 16S sequencing project. I did first a DADA2 analysis to get my taxa. Now, I would like to get a tree on these data.

I did my sequences alignment with DECIPHER. Then I try to deal with raxml-ng. The computing time is very long when I try :

./raxml-ng --msa ./output/alignement.fasta --model GTR+G --prefix test --threads 128

That's why I tried to get a consensus tree based on 100 starting trees (--start option) :

./raxml-ng --msa ./output/alignement.fasta --model GTR+G --prefix starting --tree pars{50},rand{50} --start --threads 128

./raxml-ng --consense MRE --tree starting.raxml.startTree--prefix consensus.MRE --threads 128

Then, I tried to get the bootstrap values of that tree :

./raxml-ng --bootstrap --msa ./output/alignement.fasta--prefix bootstrap --model GTR+G --seed 2 --threads 128

It is still running. I will finish by mapping these values on the consensus tree with :

raxml-ng --support --tree consensus.MRE.consensusTreeMRE --bs-trees allbootstraps --prefix bootstrap_tree --threads 32

Is that a correct way to do? Is there any parameter to trick with?

Best regards,

Paillou

Oleksiy Kozlov

unread,

Apr 3, 2024, 4:55:22 AMApr 3

to ra...@googlegroups.com

Dear Paillou,

well, consensus of random starting trees is definitely not a good substitute for the ML tree.

To make ML tree inference faster, you can specify a (taxonomic) constraint tree, and/or use the
adaptive version of raxml-ng:

https://github.com/amkozlov/raxml-ng/wiki/Installation#building-adaptive-branch

Also, please have a look at this tutorial:
https://github.com/amkozlov/ng-tutorial/wiki/evomics2024

Best,
Oleksiy

On 02.04.24 10:55, Pailler Vincent wrote:
> Dear Oleksiy,
>
> I work on a 16S sequencing project. I did first a DADA2 analysis to get my taxa. Now, I would like
> to get a tree on these data.
>
> I did my sequences alignment with DECIPHER. Then I try to deal with raxml-ng. The computing time is
> very long when I try :
>

> ./raxml-ng --msa ./output/alignement.fasta--model GTR+G--prefix test --threads 128

>
> That's why I tried to get a consensus tree based on 100 starting trees (--start option) :
>

> ./raxml-ng --msa./output/alignement.fasta--model GTR+G--prefix starting --tree pars{50},rand{50}

> --start --threads 128
>
> ./raxml-ng --consense MRE --tree starting.raxml.startTree--prefix consensus.MRE --threads 128
>
> Then, I tried to get the bootstrap values of that tree :
>
> ./raxml-ng --bootstrap --msa ./output/alignement.fasta--prefix bootstrap --model GTR+G --seed 2
> --threads 128
>
> It is still running. I will finish by mapping these values on the consensus tree with :
>
> raxml-ng --support --tree consensus.MRE.consensusTreeMRE --bs-trees allbootstraps --prefix
> bootstrap_tree --threads 32
>
> Is that a correct way to do? Is there any parameter to trick with?
>
> Best regards,
> Paillou
>

> --
> 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/b700f352-6b06-4095-8ccb-b29a1fe297b9n%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/b700f352-6b06-4095-8ccb-b29a1fe297b9n%40googlegroups.com?utm_medium=email&utm_source=footer>.

Pailler Vincent

unread,

Apr 16, 2024, 3:36:48 AMApr 16

to raxml

Dear Oleksiy,

Thanks for your feedback.

What I did is to reduce my dataset, from 19k ASV to 3.6k ASV. Also, I use the `--parse` option to optimize my number of CPUs. I used :

`./raxml-ng --log progress --threads 9 --msa ./output/alignement.fasta --model GTR+G -prefix GTR -all --bs-metric fbp,tbe --force perf_threads --bs-trees 100`

I had 20 starting trees and I also got my bootstrap values with `--all` , but with only 100 replicated. It ran well but I read a topic (https://stats.stackexchange.com/questions/86040/rule-of-thumb-for-number-of-bootstrap-samples#:~:text=Example%20(Table%20V%2C%20ibid.,95%25%20sure%2C%20850%20replications) when it is advised to use many more replicated (over 700).

It is to time-consuming if I leave the default `--bs-trees` (1000 replicates). Could I use these values on my ML tree even if my job crashes, at few hundreds of replicated?

Best,

Oleksiy Kozlov

unread,

Apr 16, 2024, 6:26:40 AMApr 16

to ra...@googlegroups.com

Dear Paillou,

raxml-ng provides an integrated bootstrap convergence test, which can be applied during the tree
inference ("--bs-trees autoMRE"), or as a post-processing step:

Please see:

https://github.com/amkozlov/raxml-ng/wiki/Tutorial#bootstrapping

https://github.com/amkozlov/raxml-ng/wiki/Tuning-model-optimization-and-tree-search#bootstrapping

Best,
Oleksiy

On 16.04.24 09:36, Pailler Vincent wrote:
> Dear Oleksiy,
>

> <https://groups.google.com/d/msgid/raxml/b700f352-6b06-4095-8ccb-b29a1fe297b9n%40googlegroups.com?utm_medium=email&utm_source=footer <https://groups.google.com/d/msgid/raxml/b700f352-6b06-4095-8ccb-b29a1fe297b9n%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/1e79310e-8945-4907-b9ca-131f0b93ccc8n%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/1e79310e-8945-4907-b9ca-131f0b93ccc8n%40googlegroups.com?utm_medium=email&utm_source=footer>.

Pailler Vincent

unread,

Apr 16, 2024, 8:42:59 AMApr 16

to raxml

Thanks again for the help, I'll try your suggestion.

Best

Paillou

Pailler Vincent

unread,

Apr 19, 2024, 9:29:45 AMApr 19

to raxml

Hi Oleksiy,

I have a .bestTree obtained with --tree pars{50},rand{50} . Is there a way to run --bs-trees autoMRE on this tree rather than on the --msa file?

Best

Oleksiy Kozlov

unread,

Apr 19, 2024, 10:40:05 AMApr 19

to ra...@googlegroups.com

autoMRE is a convergence test for bootstrap replicate trees, it does not require ML tree.

so you can use "--bootstrap --bs-trees autoMRE" (if you don't have bootstrap trees yet) or
"--bsconverge" (if you have them)

On 19.04.24 15:29, Pailler Vincent wrote:
> Hi Oleksiy,
>
> I have a .bestTree obtained with --tree pars{50},rand{50} . Is there a way to run --bs-trees autoMRE
> on this tree rather than on the --msa file?
>
> Best
>
> On Tuesday 16 April 2024 at 14:42:59 UTC+2 Pailler Vincent wrote:
>
> Thanks again for the help, I'll try your suggestion.
>
> Best
> Paillou
>
> On Tuesday 16 April 2024 at 12:26:40 UTC+2 Oleksiy Kozlov wrote:
>
> Dear Paillou,
>
> raxml-ng provides an integrated bootstrap convergence test, which can be applied during the
> tree
> inference ("--bs-trees autoMRE"), or as a post-processing step:
>
> Please see:
>
> https://github.com/amkozlov/raxml-ng/wiki/Tutorial#bootstrapping

> <https://github.com/amkozlov/raxml-ng/wiki/Tutorial#bootstrapping>
>
> https://github.com/amkozlov/raxml-ng/wiki/Tuning-model-optimization-and-tree-search#bootstrapping <https://github.com/amkozlov/raxml-ng/wiki/Tuning-model-optimization-and-tree-search#bootstrapping>

>
> Best,
> Oleksiy
>
> On 16.04.24 09:36, Pailler Vincent wrote:
> > Dear Oleksiy,
> >
> > Thanks for your feedback.
> >
> > What I did is to reduce my dataset, from 19k ASV to 3.6k ASV. Also, I use the `--parse`
> option to
> > optimize my number of CPUs. I used :
> >
> > `./raxml-ng --log progress --threads 9 --msa ./output/alignement.fasta --model GTR+G
> -prefix GTR
> > -all --bs-metric fbp,tbe --force perf_threads --bs-trees 100`
> >
> > I had 20 starting trees and I also got my bootstrap values with `--all` , but with only 100
> > replicated. It ran well but I read a topic
> >

> (https://stats.stackexchange.com/questions/86040/rule-of-thumb-for-number-of-bootstrap-samples#:~:text=Example%20(Table%20V%2C%20ibid <https://stats.stackexchange.com/questions/86040/rule-of-thumb-for-number-of-bootstrap-samples#:~:text=Example%20(Table%20V%2C%20ibid>.,95%25%20sure%2C%20850%20replications) when it is advised to use many more replicated (over 700).

> <https://groups.google.com/d/msgid/raxml/b700f352-6b06-4095-8ccb-b29a1fe297b9n%40googlegroups.com?utm_medium=email&utm_source=footer <https://groups.google.com/d/msgid/raxml/b700f352-6b06-4095-8ccb-b29a1fe297b9n%40googlegroups.com?utm_medium=email&utm_source=footer> <https://groups.google.com/d/msgid/raxml/b700f352-6b06-4095-8ccb-b29a1fe297b9n%40googlegroups.com?utm_medium=email&utm_source=footer <https://groups.google.com/d/msgid/raxml/b700f352-6b06-4095-8ccb-b29a1fe297b9n%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/1e79310e-8945-4907-b9ca-131f0b93ccc8n%40googlegroups.com <https://groups.google.com/d/msgid/raxml/1e79310e-8945-4907-b9ca-131f0b93ccc8n%40googlegroups.com>
> >

> <https://groups.google.com/d/msgid/raxml/1e79310e-8945-4907-b9ca-131f0b93ccc8n%40googlegroups.com?utm_medium=email&utm_source=footer <https://groups.google.com/d/msgid/raxml/1e79310e-8945-4907-b9ca-131f0b93ccc8n%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/0498ee9e-f04b-4a66-aa52-4660168c41ban%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/0498ee9e-f04b-4a66-aa52-4660168c41ban%40googlegroups.com?utm_medium=email&utm_source=footer>.

Pailler Vincent

unread,

Apr 19, 2024, 10:48:30 AMApr 19

to ra...@googlegroups.com

Thanks

I run the MRE bootstrap with the msa file then. I will use --support to add the values on my best tree.

Best

To unsubscribe from this group and stop receiving emails from it, send an email to raxml+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/raxml/50c0ba86-2b33-4ea5-9622-2b739473c7ca%40gmail.com.

Pailler Vincent

unread,

May 2, 2024, 11:40:15 AMMay 2

to raxml

Hi, my job crashed after 7 days with 270 replicates. Is it OK to run 4 jobs in parallel, on the same matrix, to get 270 replicates * 4 jobs and then, use --bsconverge to converge all of that?

Best

Oleksiy Kozlov

unread,

May 2, 2024, 2:25:32 PMMay 2

to ra...@googlegroups.com

yes it's OK, just make sure you specify distinct --seed for each job!

> https://github.com/amkozlov/raxml-ng/wiki/Tuning-model-optimization-and-tree-search#bootstrapping <https://github.com/amkozlov/raxml-ng/wiki/Tuning-model-optimization-and-tree-search#bootstrapping> <https://github.com/amkozlov/raxml-ng/wiki/Tuning-model-optimization-and-tree-search#bootstrapping <https://github.com/amkozlov/raxml-ng/wiki/Tuning-model-optimization-and-tree-search#bootstrapping>>

> >
> > Best,
> > Oleksiy
> >
> > On 16.04.24 09:36, Pailler Vincent wrote:
> > > Dear Oleksiy,
> > >
> > > Thanks for your feedback.
> > >
> > > What I did is to reduce my dataset, from 19k ASV to 3.6k ASV. Also, I use the
> `--parse`
> > option to
> > > optimize my number of CPUs. I used :
> > >
> > > `./raxml-ng --log progress --threads 9 --msa ./output/alignement.fasta
> --model GTR+G
> > -prefix GTR
> > > -all --bs-metric fbp,tbe --force perf_threads --bs-trees 100`
> > >
> > > I had 20 starting trees and I also got my bootstrap values with `--all` , but
> with only 100
> > > replicated. It ran well but I read a topic
> > >
> >

> (https://stats.stackexchange.com/questions/86040/rule-of-thumb-for-number-of-bootstrap-samples#:~:text=Example%20(Table%20V%2C%20ibid <https://stats.stackexchange.com/questions/86040/rule-of-thumb-for-number-of-bootstrap-samples#:~:text=Example%20(Table%20V%2C%20ibid> <https://stats.stackexchange.com/questions/86040/rule-of-thumb-for-number-of-bootstrap-samples#:~:text=Example%20(Table%20V%2C%20ibid <https://stats.stackexchange.com/questions/86040/rule-of-thumb-for-number-of-bootstrap-samples#:~:text=Example%20(Table%20V%2C%20ibid>>.,95%25%20sure%2C%20850%20replications) when it is advised to use many more replicated (over 700).

> > >
> > > --
> > > 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
> > >
> >

> >
> > --
> > 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/0498ee9e-f04b-4a66-aa52-4660168c41ban%40googlegroups.com <https://groups.google.com/d/msgid/raxml/0498ee9e-f04b-4a66-aa52-4660168c41ban%40googlegroups.com>
> >

> <https://groups.google.com/d/msgid/raxml/0498ee9e-f04b-4a66-aa52-4660168c41ban%40googlegroups.com?utm_medium=email&utm_source=footer <https://groups.google.com/d/msgid/raxml/0498ee9e-f04b-4a66-aa52-4660168c41ban%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.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/raxml/50c0ba86-2b33-4ea5-9622-2b739473c7ca%40gmail.com

> <https://groups.google.com/d/msgid/raxml/50c0ba86-2b33-4ea5-9622-2b739473c7ca%40gmail.com>.

>
> --
> 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/8aefc758-1a88-4604-a23c-98def0f9d6a7n%40googlegroups.com
> <https://groups.google.com/d/msgid/raxml/8aefc758-1a88-4604-a23c-98def0f9d6a7n%40googlegroups.com?utm_medium=email&utm_source=footer>.

Reply all

Reply to author

Forward